Avifauna of an Early Successional Habitat Along the Middle Missouri River

Introduction

Woodland habitats within the northern Great Plains are primarily restricted to river corridors, with the Missouri River corridor being the most prominent riparian corridor in this region. Large-scale historical changes in the riparian vegetation of the Missouri River corridor have occurred and continue to the present. Prominent among these changes is the conversion of riparian woodland to agricultural lands, with estimates of 40 to 80 percent reductions in forested area in different stretches of the middle Missouri River (Hesse et al. 1988, Hesse 1996). Construction of dams along the river has created large, main-stem reservoirs that have flooded large expanses of riparian habitat, further reducing the availability of riparian vegetation. In addition, with the flood control provided by the dams and the channelization of the riverbed, periodic flooding along the middle Missouri River has been all but eliminated. Thus, early successional habitats that invade flood-scoured areas have become increasingly rare along the middle Missouri River (Hesse 1996). Furthermore, with the elimination of periodic flooding, remaining riparian forests have matured, which results in a higher proportion of mature riparian forests relative to other successional stages (Hesse 1996). As a result of these historical changes in Missouri River floodplain use and management, early successional habitats are perhaps the most reduced riparian habitat type along the middle Missouri River. Woody vegetation of early successional habitats along the middle Missouri River is dominated by willow (Salix spp), young cottonwood (Populus deltoides), and dogwood (Cornus spp). These species are superceded in more mature riparian forests of southeastern South Dakota by larger cottonwoods, American elm (Ulmus americana), green ash (Fraxinus pennsylvanica), hackberry (Celtis occidentalis), box elder (Acer negundo), silver maple (Acer saccharinum), bur oak (Quercus macrocarpa), black walnut (Juglans nigra), and basswood (Tilia americana) (Hesse 1996).

Few quantitative data exist that describe the breeding avifauna of riparian woodland habitats in the northern Great Plains and along the Missouri River in particular. The relationship between vegetational succession and breeding bird abundance and/or richness in various forested habitat types in North America has been examined in several studies. Species richness, in general, tends to be higher in later successional stages due to higher vegetational and structural diversity (Shugart and James 1973, Zimmerman and Tatschl 1975, Schieck et al. 1995, Buffington et al. 1997). However, in some cases early successional habitats have been shown to have higher richness than mid-successional (Schieck et al. 1995) or climax (Karr 1968) stages. Avian abundances are also higher in late than in early successional stages in some habitats (Karr 1968, Swift et al. 1984, Schieck et al. 1995), but no difference or the reverse has also been reported (Zimmerman and Tatschl 1975, Probst 1979, Anderson et al. 1983, Morgan and Freedman 1986, Liknes et al. 1994). The precise nature of the relationship between vegetational succession and avian abundance and species richness in riparian woodlands of the middle Missouri River is unknown.

Some birds, such as Bell's vireo (Vireo bellii) and American redstart (Setophaga ruticilla), are prominently associated with scrubby, early successional habitats in South Dakota and in other portions of their range (Brown 1993, Peterson 1995, Hunt 1996). Further, reductions in available early successional habitats are associated with population declines in several Neotropical migrant bird species, including American redstart, in other regions of the country (Litvaitis 1993), but the possibility for such a relationship has not been studied in the northern Great Plains. Declining early successional habitats along the middle Missouri River may serve as critical breeding habitat for a variety of bird species. As a result, such habitats may merit special conservation attention, but the occurrence of birds in these habitats during the breeding season has not been quantified. Thus, my study examines the abundance and species composition of the bird community of an early successional habitat along the Missouri River in southeastern South Dakota during the breeding season. Fixed-radius point counts and mist-net sampling were used to document abundance, species composition, and breeding status of birds within this habitat during breeding seasons from 1994 through 1996.

Study Area

The study site was an early successional riparian habitat along the Missouri River in Union County, South Dakota (42° 42' N, 96° 48' W). The early successional habitat at this site ran in a relatively narrow strip (approximately 75 m wide by 900 m long) adjacent and parallel to the river. Vegetation in the central 500 m of this strip was dominated by dense willow stands, consisting primarily of saplings (mean canopy height = 3 m), interspersed with clumps of dogwood, green ash, false indigo (Amorpha fruticosa), and sumac (Rhus spp.) (Martin 1996). This willow-dominated habitat was flanked at both ends by shrubby, dogwood-dominated vegetation, with some willows and a few scattered cottonwood or ash trees, that extended for approximately 200 m in both directions. The early successional habitat was bordered at either end and by a narrow strip on the side farthest removed from the river by more mature forest with cottonwood, elm, and ash trees of at least 10 m in height and forming a nearly closed canopy.

Methods

Both point counts and mist netting have biases as survey methods for documenting avian abundance and species composition. When used in isolation, both of these survey methods produce incomplete sampling of the avian community (Bibby et al. 1992, Remsen and Good 1996). However, conducting both point counts and mist netting in the same habitat can reduce these biases because biases of one method can be somewhat offset by the other method (Rappole et al. 1998). Both fixed-radius (25 m) point counts and mist netting were used to sample the avian community in an early successional habitat along the middle Missouri River.

Five points were established in the early successional habitat along a roughly linear transect paralleling the Missouri River. Points were separated by 200 m, so the total distance covered by the transect was 800 m. The three middle points along the transect were located in the willow-dominated vegetation. The two points at either end of the transect penetrated the dogwood-dominated habitat with some willow saplings and a few scattered cottonwood and ash trees present. Surveys were conducted five times during each of the three breeding seasons and survey dates included early (3) and late (17 to 21) June, early (8 to 14) and late (23 to 25) July, and early (4 to 6) August. All counts were conducted between 0553 and 0916 CST and counts were not conducted on days with rain or winds greater than 35 km per hour. Within each breeding season, successive surveys were separated by at least 10 days and the direction in which the transect was sampled was reversed on successive surveys to reduce possible temporal bias. The number of points and number of replicates used have been shown to provide stable density estimates in habitats with heterogeneous vegetation (Morrison et al. 1981). All birds detected by sight or by sound were identified and counted and their distance from the point center was measured with a Ranging Model 620 rangefinder. Distances were recorded as inside or outside 25 m from the point center (Hutto et al. 1986, Bibby et al. 1992).

Survey periods lasted 10 min per point. Birds detected while I was walking between points were counted and their distance from the nearest point recorded. Birds detected flying overhead were counted only if they potentially used the early successional habitat. For example, great blue herons (Ardea herodias), least terns (Sterna antillarum), and bank swallows (Riparia riparia) were not counted. Overall abundance for the three years pooled for all birds and for individual species was calculated from detections inside 25 m to calculate densities (birds km^-2) and from all detections (inside and outside 25 m) to calculate relative abundances (birds/point).

A total of two to five mist nets (2.6 m by 9 m, 30 mm mesh) was erected on each sampling date. Sampling dates occurred from 3 June to 4 August in 1994, 3 June to 7 August in 1995, and 3 June to 6 August in 1996. The number of dates sampled by mist netting was six in 1994, nine in 1995, and seven in 1996. All nets were erected within the willow-dominated habitat and away from the boundary with the more mature forest by at least 10 m. The same net locations were sampled every year, although net placement usually varied among sampling dates within a single breeding season. Nets were erected within 1.5 hr of sunrise and were left open for at least 2 hr, but were always closed by 1030 CST. Mist net sampling was not conducted on days with rain or winds in excess of 35 km per hour. Audiotapes of eastern screech-owl (Otus asio) calls of 30 min duration were played at net sites to draw birds into the nets and tapes were systematically rotated among nets. For captured birds, measurements of mass (to the nearest 0.1 g on an Ohaus Model LS 200 electronic balance), unflattened wing chord (to the nearest 0.1 mm), and visible subcutaneous fat score, on a scale of 0 to 5 according to Helms and Drury (1960), were obtained. Breeding condition of captured birds was assessed by whether or not a cloacal protuberance or brood patch was present. The presence of a well-developed brood patch in female passerines is often considered physiological evidence for breeding (e.g., Peterson 1995). I also determined sex of captured passerine birds by the presence of a cloacal protuberance or brood patch (Pyle et al. 1987). Captured birds were aged, when possible, by skull pneumatization (Pyle et al. 1987). Following capture, birds were banded, under Federal Banding Permit 22199, with a standard U.S. Fish and Wildlife Service aluminum leg band and released.

Results

The total number of bird species detected on point counts over all three years combined was 44 (Table 1), 40 of which are documented breeders in riparian habitats of southeastern South Dakota (SDOU 1991). All potential breeding species in this habitat that were captured by mist net, except for mourning doves (Zenaida macroura) and hairy woodpeckers (Picoides villosus), exhibited physiological evidence for breeding, cloacal protuberances or brood patches, or were captured as juveniles, strongly suggesting breeding of these species (n = 26, Table 2) in this or nearby habitats. The annual numbers of species detected on point counts were 33 in 1994, 36 in 1995, and 35 in 1996. The mean number of species detected on point counts per sampling date was 22.3 ± 2.3 (SD) in 1994, 22.6 ± 3.7 in 1995, and 21.0 ± 3.2 in 1996. These annual means did not differ significantly (one-way ANOVA, F_2,12 = 0.392, P = 0.68), which indicates no annual variation in species richness detected by point counts.

A total of 227 net hrs (1 net hr = 1 net open for 1 hr) was generated over the three breeding seasons, with annual net hr totals of 60 in 1994, 74 in 1995, and 93 in 1996. The percentages of total net hrs for which owl tapes were played were 16.9% in 1994, 18.9% in 1995, and 12.9% in 1996, for an overall average of 15.9%. Over all three years combined, 31 species of birds were captured, 28 of which were potential breeding species (Table 2). Total annual species captures were 22 in 1994, 20 in 1995, and 15 in 1996. Combining both point count and mist net data, the total number of bird species detected over the three-year study period was 49, 44 of which are breeding species in southeastern South Dakota riparian habitats (SDOU 1991). Annual totals for species detected by both methods were 37 in 1994, 40 in 1995, and 37 in 1996. Of the 49 species detected, 26 were detected by both point count and mist net sampling methods, 18 by point counts only, and 5 by mist net only. All species detected by mist net only (Table 2) were represented by a single capture. For species detected only on point counts, most were also rare in the early successional habitat at my study site, with densities below 20 birds km^-2 and/or relative abundances below 0.2 birds/point. Exceptions included eastern kingbird (Tyrannus tyrannus, 122.2 birds km^-2, 0.68 birds/point), European starling (Sturnus vulgaris, 34.0 birds km^-2, 0.39 birds/point), common grackle (Quiscalus quiscula, 47.5 birds km^-2, 0.19 birds/point), song sparrow (Melospiza melodia, 47.5 birds km^-2, 0.17 birds/point), red-winged blackbird (Agelaius phoeniceus, 27.2 birds km^-2, 0.15 birds/point), and red-bellied woodpecker (Melanerpes carolinus, 27.2 birds km^-2, 0.11 birds/point).

No significant variation in relative abundance over the breeding season was apparent. Mean relative abundance for survey periods ranged from 14.7 ± 2.1 (SD) birds/point in early August to 19.7 ± 1.5 birds/point in early July, but these values did not differ significantly from each other or from those during other survey periods (one-way ANOVA, F_4,10 = 1.50, P = 0.27). Overall density for all birds over all three years was 4,033.9 birds km^-2 and overall relative abundance was 16.3 birds/point. Annual densities and relative abundances were 4,074.4 birds km^-2 and 16.7 birds/point in 1994, 4,563.3 birds km^-2 and 17.4 birds/point in 1995, and 3,463.2 birds km^-2 and 14.8 birds/point in 1996. Density and relative abundance calculations were generally in agreement with regard to abundances of individual species and American goldfinch (Carduelis tristis), yellow warbler (Dendroica petechia), gray catbird (Dumetella carolinensis), and Bell's vireo were the most abundant species by both calculations (Table 1).

For all three years pooled, the overall capture rate for all birds was 115.0 birds/100 net hr and annual overall capture rates were 165.0 birds/100 net hr in 1994, 163.5 birds/100 net hr in 1995, and 44.1 birds/100 net hr in 1996. The species with the highest capture rate was the yellow warbler at 21.6 birds/100 net hr. Other species with capture rates greater than 10 birds/100 net hr included Bell's vireo, American goldfinch, and gray catbird (Table 2).

Discussion

Point counts and mist net sampling both have biases as sampling methods for avian communities (Bibby et al. 1992, Remsen and Good 1996, Rappole et al. 1998). Remsen and Good (1996) argued that constant-effort mist net sampling protocols do not allow comparison of relative abundances of species from capture data because of many confounding variables such as differences among species in net avoidance, proportion of territorial individuals and floaters, vertical distributions within the habitat, and flight frequencies or distances. Point counts also have some drawbacks as measures of abundance because they tend to miss or underestimate certain species, particularly those with low densities or secretive, skulking habits (Bibby et al. 1992, Rappole et al. 1998). Spot mapping techniques are arguably considered the most accurate measure of avian density, but they are time and labor intensive relative to point counts (Bibby et al. 1992, Dobkin and Rich 1998). Several studies have compared spot mapping and point counts as measures of avian species richness, relative abundance, and density (DeSante 1981, 1986, Szaro and Jakle 1982, Dobkin and Rich 1998). Generally, these studies have found that point counts produce errors in actual density estimates, but that with repeated visits, point counts provide acceptable measures of species richness and relative abundance, with lower effort than spot mapping. Dobkin and Rich (1998) recommended fixed-radius point counts with at least two visits per site for breeding season surveys and their study, as well as that of Szaro and Jakle (1982), included willow-dominated habitats similar to those in my study. Finally, Rappole et al. (1998) found that weaknesses of point count and mist net sampling methods tended to be offset by strengths of the other method, so that a combination of the two methods provided a more reliable assessment of the avian community than either method did alone.

In my study, abundances were calculated from point count detections in two different ways. First, densities (birds km^-2) were calculated from all detections of 25 m or less from the observer. Such measurement of density is standard for point counts (Reynolds et al. 1980) but, because birds move into the count area from outside during the count period, the effective distance sampled is unknown and therefore, absolute density cannot be accurately calculated (Hutto et al. 1986). Thus, these density estimates are in reality another measure of relative abundance. The second abundance calculation in my study, which I termed relative abundance (birds/point), was calculated from all detections, regardless of their distance from the observer. Both abundance measures in my study were in general agreement, at least for the more common species (Table 1). Indeed, the top-four species were ranked in the same order by both methods and 18 of the top-20 ranked species were the same, although individual species ranks were not in the same order. However, rank order never differed by more than five places for any of the top-20 ranked species.

Because of differences among species in territory size, proportion of floaters, vertical distributions within the habitat, flight frequencies or distances, and response to taped owl calls, capture rates cannot effectively be used as a measure of relative abundance (Remsen and Good 1996). Nevertheless, the four most common species on point counts in my study were also the four most commonly captured species, although their rank order differed between the two methods. Rankings of individual species beyond these top-four, however, sometimes differed greatly between point counts and mist net sampling. For example, downy woodpeckers (Picoides pubescens), black-capped chickadees (Poecile atricapillus), ovenbirds (Seiurus aurocapillus), and common yellowthroats (Geothlypis trichas) were more commonly captured than expected based on their abundance from point counts. Low abundance on point counts relative to capture rates may be due to secretive or skulking habits, in the case of ovenbirds and common yellowthroats, or to ready responses of these species to taped owl calls (pers. obser.). Several species that were not captured in mist nets were relatively common on point counts. These included red-bellied woodpeckers, eastern kingbirds, song sparrows, common grackles, and red-winged blackbirds. Several other species were also rarely captured, although they were relatively common on point counts. These species included eastern wood-pewee (Contopus virens), cedar waxwing (Bombycilla cedrorum), indigo bunting (Passerina cyanea), brown-headed cowbird (Molothrus ater), orchard oriole (Icterus spurius), and Baltimore oriole (Icterus galbula). These rarely captured species tend to remain high in the canopy or are associated with adjacent marshy or forested habitats that were not sampled with mist nets in my study.

Differences in survey methods or protocols, in addition to differences in avian detectability among different habitat types make comparisons of species richness and abundance among different studies and locations difficult at best. Nevertheless, coarse comparisons are possible and such comparisons suggest relatively high avian species richness and abundance in the early successional habitat in my study. Overall density and relative abundance for all birds at my study site were 4,033.9 birds km^-2 and 16.3 birds/point, respectively. This density is greater than those detected by Liknes et al. (1994) during a single-year study in which they used similar techniques in cottonwood/ash/elm (1,169.1 birds km^-2) and cottonwood/willow/dogwood (1,782.5 birds km^-2) habitats along the Missouri River in southeastern South Dakota. The 37 to 40 species detected annually in early successional habitats is also greater than the 25 and 31 species detected by Liknes et al. (1994) for the cottonwood/ash/elm and cottonwood/willow/dogwood habitats, respectively. Although differences in methods confound direct comparison, overall avian species richness and abundance in my study compare favorably to species richness and/or abundance measures from Kansas floodplain forests (Zimmerman and Tatschl 1975) and from different successional forest stages in Illinois (Karr 1968), Arkansas (Shugart and James 1973), South Carolina (Buffington et al. 1997), Massachusetts (Swift et al. 1984), and Nova Scotia (Morgan and Freedman 1986).

Zimmerman and Tatschl (1975) studied floodplain forests of the Missouri River near Fort Leavenworth, Kansas and found that avian abundance was higher in young (20 years-old) forest than in mature forest and old field habitats, while avian diversity was highest in the mature forest. Liknes et al. (1994) noted lower avian abundance and species richness in a mature forest than in an earlier successional stage along the Missouri River in southeastern South Dakota. These studies, along with the current study, suggest the interesting possibility that avian abundance, and perhaps species richness, may be higher in early successional stages than in more mature forests along the middle Missouri River. The relationship between vegetational succession and avian species richness and abundance along the middle Missouri River merits further investigation, particularly in light of the declining availability of early successional habitats (Hesse 1996).

Of the 44 breeding species detected in my study by point counts and mist netting, 20 (45.5%) were Neotropical migrants, 12 (27.3%) were temperate-zone migrants, and 12 (27.3%) were permanent residents in South Dakota (SDOU 1991, DeGraaf and Rappole 1995). This percentage of Neotropical migrants is similar to that in other Missouri River riparian habitats (50 to 53%, Liknes et al. 1994) and in other deciduous woodlands in the northern Great Plains (49%, Faanes 1984) and central North America (45%, Terborgh 1989). The percentage of Neotropical migrants in my study is lower than the percentage documented for avian communities of eastern deciduous forests or Rocky Mountain riparian forests (68 to 70%), but is greater than the percentage from riparian habitats in California (25%) or Rocky Mountain coniferous forests (18%, Terborgh 1989).

The only species detected in my study that is monitored by the South Dakota Natural Heritage Program (Stukel and Backlund 1997) was the black-and-white warbler (Mniotilta varia) that was captured in a mist net on 29 July 1994. This was a juvenile bird (determined by degree of skull ossification, Pyle et al. 1987) that was probably an early fall migrant as this was the only time that this species was observed during my study. However, the possibility of breeding by black-and-white warblers my study area or nearby habitats cannot be ruled out because this bird had low visible subcutaneous fat scores, which are generally characteristic of, although not exclusive to, nonmigratory individuals, and this species has been reported as nesting along the Missouri River in other parts of South Dakota (SDOU 1991) and Nebraska (Johnsgard 1979).

The early successional habitat in my study was occupied by a wide variety of birds, including a number of Neotropical migrants that have exhibited population declines in central North America or throughout their range (DeGraaf and Rappole 1995), such as yellow-billed cuckoo (Coccyzus americanus), eastern wood-pewee, Bell's vireo, ovenbird, common yellowthroat, and orchard oriole. Moreover, Bell's vireo, American redstart, and ovenbird are rare or local in southeastern South Dakota (SDOU 1991). Bell's vireos occupy early successional habitats in riparian areas throughout their range (Brown 1993) and have previously been shown to nest in my study area at relatively high density (Martin 1996). American redstarts have also been shown to occupy early successional habitats in high density in other portions of their range (Hunt 1996) and these habitats appear to serve as source habitats for this species (Hunt 1998).

Acknowledgments

I thank Eric Liknes, Kurt Dean, Johnida Martin, Mark Drymalski, and Jess Brown for their assistance with various aspects of my project. I also thank Doug Backlund and Eileen Dowd Stukel and two anonymous reviewers for constructive comments on an earlier version of my manuscript.

Literature Cited