1149
Songbird diversity and movement in upland and riparian habitats in the boreal mixedwood forest of northeastern Ontario Erin Mosley, Stephen B. Holmes, and Erica Nol
Abstract: Little is known about the importance of riparian areas in supporting avifaunal diversity in the boreal mixedwood forest, especially outside of the breeding season. Bird populations were sampled by mist netting 18 upland and 18 riparian sites along six streams in a forested region of northeastern Ontario. Riparian sites generally had more variable vegetation than upland sites. Some riparian sites formed distinctive habitats, while others were structurally and compositionally similar to upland sites. During spring and fall migration, there was no significant difference in bird abundance or species richness between riparian and upland habitats. During the breeding period, riparian areas had greater avian species richness and abundance and more insects than upland forests, suggesting that birds were selecting these habitats because they contain more food. More birds were captured in nets placed perpendicular to the stream than parallel during the breeding and fall migration periods, suggesting that riparian areas may function as movement corridors. A greater understanding of the importance of riparian habitats to songbird communities is needed if we are to maximize the effectiveness of these regions for conserving avian biodiversity in the boreal mixedwood forest. Résumé : On sait peut de choses de l’importance des milieux riverains pour le maintien de la diversité de l’avifaune dans la forêt boréale mixte, en particulier en dehors de la saison de reproduction. Les populations d’oiseaux ont été échantillonnées par capture au filet japonais à 18 sites non riverains et 18 sites riverains de six ruisseaux dans une région boisée du nord-est de l’Ontario. Les sites riverains avaient en général une végétation plus variable que les sites non riverains. Certains sites riverains constituaient des habitats distincts tandis que d’autres avaient une structure et une composition semblables à celles des sites non riverains. Durant les migrations de printemps et d’automne, il n’y avait pas de différence significative dans l’abondance ou la richesse spécifique des oiseaux entre les milieux riverains et non riverains. Durant la pariade, la richesse spécifique en oiseaux de même que l’abondance d’oiseaux et d’insectes étaient plus élevées dans les milieux riverains que non riverains, ce qui suggère que les oiseaux choisissaient ces habitats parce qu’ils contiennent plus de nourriture. Durant la pariade et la migration automnale, plus d’oiseaux ont été capturés dans les filets placés perpendiculairement aux ruisseaux que dans ceux placés parallèlement, ce qui suggère que les milieux riverains constituent des corridors de déplacement. Une meilleure compréhension de l’importance des milieux riverains pour les passereaux est nécessaire si on veut maximiser la contribution de ces milieux à la conservation de la biodiversité aviaire de la forêt boréale mixte. [Traduit par la Rédaction]
Mosley et al.
1164
Introduction Riparian habitats often support greater species richness than other adjacent sites (McGarigal and McComb 1992; Murray and Stauffer 1995; Naiman and Decamps 1997; Bub et al. 2004), in part because many terrestrial species are dependent on riparian habitats at some stage of their life history (Naiman et al. 1993; LaRue et al. 1995). Several factors Received 16 July 2005. Accepted 9 January 2006. Published on the NRC Research Press Web site at http://cjfr.nrc.ca on 19 April 2006. E. Mosley1,2 and E. Nol. Department of Biology, Trent University, 1600 West Bank Drive, Peterborough, ON K9J 7B8, Canada. S.B. Holmes. Great Lakes Forestry Centre, Canadian Forest Service, Natural Resources Canada, 1219 Queen Street E, Sault Ste. Marie, ON P6A 2E5, Canada. 1 2
Corresponding author (e-mail:
[email protected]). Present address: Ontario Parks, 435 James Street S, Suite 221d, Thunder Bay, ON P7E 6S8, Canada.
Can. J. For. Res. 36: 1149–1164 (2006)
have been used to explain the value of riparian areas to animals, including greater structural diversity of vegetation, higher primary and secondary productivity related to nutrient inputs, the presence of an edge ecotone, and increased horizontal diversity resulting from the presence of water, riparian, and upland ecosystems as well as frequent flooding, which brings in nutrients and creates regeneration niches (LaRue et al. 1995; Boyce and Payne 1997; Naiman and Decamps 1997; Brinson and Verhoeven 1999). This increased diversity and the sedimentation filter properties of riparian zones, which are important in maintaining water quality and fish habitat, are used to justify the maintenance of riparian buffers during forest harvesting operations (Darveau et al. 1995; LaRue et al. 1995; Whitaker and Montevecchi 1997). Current forest management guidelines in Ontario, Canada, recommend that riparian reserves 30 to 90 m wide be maintained adjacent to lakes and streams, depending on the slope of the surrounding shoreline (Ontario Ministry of Natural Resources, hereafter OMNR, 1988). These guidelines were developed to protect fish habitat and water quality (OMNR 1988), although songbirds may also
doi:10.1139/X06-010
© 2006 NRC Canada
1150
benefit through the principle that links habitat structural diversity with avian diversity (MacArthur and MacArthur 1961; Karr and Roth 1971; LaRue et al. 1995; Murray and Stauffer 1995; Whitaker and Montevecchi 1997). Previous research comparing songbird use of upland and riparian habitats has generated mixed conclusions, especially when the vegetation structure is similar between the two habitat types (Bub et al. 2004). Some studies have found similarities in bird abundance and species richness in riparian and upland areas (Murray and Stauffer 1995; Whitaker and Montevecchi 1997), others have found riparian areas to support a greater abundance and (or) diversity of birds than nonriparian areas (LaRue et al. 1995; Bub et al. 2004), while still others have found riparian areas to support lower bird abundance and species richness than upland areas (McGarigal and McComb 1992). Research on the importance of riparian areas in supporting avifaunal diversity and abundance in the boreal mixedwood forest is limited, especially outside of the breeding season. We compare bird species diversity in riparian and adjacent upland areas during the migratory and breeding seasons in northeastern Ontario, Canada. Our objective was to measure whether riparian or upland areas support higher avian abundance and species richness, which we tested by (1) assessing an index of bird abundance and species richness in riparian and upland habitats, (2) assessing whether vegetation composition and structure differ between these two habitat types, and (3) providing an assessment of insect prey availability in riparian and upland areas, as a measure of food availability, to explain potential differences in bird communities.
Materials and methods Study area and design The study was conducted near the town of White River (48°35′N, 85°15′W) in the boreal mixedwood forest of northeastern Ontario, Canada (Fig. 1). Six streams in three different watersheds were selected in large forest stands managed by Domtar Inc. Stands were selected to represent a range of overstory dominance conditions, representing typical mature forest stands in the boreal mixedwood ecosystem. At the time of the study, about 10% of the surrounding landscape had been clear-cut, although very little of this was close to the sampling blocks. The distance to the nearest cut block ranged from 200 to 3800 m (median 2700 m). The forest is a mosaic of trembling aspen (Populus tremuloides Michx.), white birch (Betula papyrifera Marsh.), balsam fir (Abies balsamea (L.) Mill.), and black spruce (Picea mariana (Mill.) BSP) in upland habitats, and white birch, balsam fir, black spruce, and white spruce (Picea glauca (Moench) Voss) in riparian habitats. Shrub species consisted primarily of alder (green alder (Alnus crispa (Ait.) Pursh) and speckled alder (Alnus rugosa (Du Roi) Spreng.)), beaked hazel (Corylus cornuta Marsh.), Labrador tea (Ledum groenlandicum Oeder), leatherleaf (Chamaedaphne calyculata (L.) Moench), sweet gale (Myrica gale L.), mountain maple (Acer spicatum Lam.), mountain ash (Sorbus americana Marsh.), and serviceberry (Amelanchier spp.). Stands ranged in age from 52 to 95 years and had not been logged previously.
Can. J. For. Res. Vol. 36, 2006
Songbird sampling Mist netting was used to sample bird populations. Twelve mist nets (12 m × 2.6 m, 30 mm mesh size; DeSante et al. 2002) were used in each of the six sampling blocks. Two nets were established at each of the three riparian and three upland sampling sites per block. In the riparian areas, one net in each pair was established parallel to the stream, and the other was perpendicular to the stream, to assess movement patterns within the riparian areas. Nets were perpendicular to one another in upland areas, but without reference to any landmark (Fig. 2). Birds were captured in 2003 during spring migration (14 May – 19 June), the breeding and juvenile dispersal period (24 June – 15 August), and fall migration (26 August – 2 October). During each time period, each site was sampled three times. Nets were opened for 6 consecutive hours per day, beginning 30 min before sunrise. Captured birds were identified to species, banded with a numbered aluminum Canadian Wildlife Service band, aged, and sexed before release. Indices of abundance and bird species richness were calculated using the total number of captures per site in each sampling period. If a bird was recaptured during the same sampling period, only the original capture was included in the analysis. Insect sampling Flying insects were sampled as an index of prey availability using two colourless, translucent plastic double-sided sticky trap boards (26 cm × 26 cm coated with Tanglefoot®), fitted together at their midpoints to form intersecting planes. One trap set was placed at each sampling site at a height of 2.6 m (the height of the mist nets), for a total of 36 boards per habitat type. Collections were made at 21-day intervals from 14 May to 30 September 2003 for a total of six samples, which corresponded to the period of bird sampling. Insect samples that were destroyed by black bears (Ursus americanus) or wind were excluded from analyses, leaving a total of 165 samples. After collection, traps were disassembled, and each board was covered in plastic wrap and refrigerated at 6 °C. Insects were identified to order and were sorted into size classes based on measured length (15 mm). The counts and measurements were used to quantify indices of abundance and biomass of aerial invertebrates for each site. Insects 10 cm diameter at breast height (DBH)) were identified to species and enumerated, and their DBHs were measured (2 cm increments) to calculate basal area (m2/ha) by species (Whitaker and Montevecchi 1997). Standing dead © 2006 NRC Canada
Mosley et al.
1151
Fig. 1. The study took place near the town of White River (48°35′N, 85°15′W), northeastern Ontario, Canada. The image on the right illustrates the proximity of study blocks to one another.
Fig. 2. Diagrammatic representation of sampling block, illustrating location of mist nets (black rectangle), sticky traps (white square), and vegetation sampling plots (gray circle). An example of the individual sampling site is illustrated with a hatched circle.
© 2006 NRC Canada
1152
trees were enumerated by decay class (modified from Watt and Caceres 1999) and DBH class. Two 11.3 m transects were established in each plot to measure shrub (wood species 0.5 m tall) density by species. Along the transect all shrubs that made contact with a 2 m rod, held at breast height, were enumerated (Murray and Stauffer 1995). Percent cover of herbaceous plants by species was measured in two 1 m × 1 m plots located 5 m due north and due south of the centre of each vegetation plot. Canopy height was measured for two randomly selected trees using a clinometer and then averaged. The vertical structure of the vegetation was assessed at four layers: low shrub (0–0.5 m), high shrub (>0.5–2 m), sapling (>2–5 m), and canopy (>5 m). For each layer, percent cover was estimated by two observers and averaged. Each site was classified into one of the 29 forest ecosystem classification (FEC) ecosites and 28 vegetation types using the key from the FEC field guide of northeastern Ontario (Welsh and Lougheed 1996; Taylor et al. 2000). Statistical analyses ANOVA models, t tests, and nonparametric tests were run using SPSS version 11.0 (SPSS Inc., Chicago, Illinois). Data were tested for normality using Kolmogorov–Smirnov tests and for equal variances using Levene’s tests (Zar 1999) and log transformed if necessary. If the assumptions of parametric tests were still not met, nonparametric tests were used. Because many of the tests had relatively low power because of small sample sizes, alpha levels were set at 0.10 (Sokal and Rohlf 1995; Rodewald and Abrams 2002). Sequential Bonferroni corrections were not used, as recent literature suggests that they are overly conservative (e.g., Moran 2003; Garcia 2004; Neuhäuser 2004). Principal components analysis (PCA), redundancy analysis (RDA), and detrended correspondence analysis (DCA) were run on untransformed data using CANOCO version 4.5 (Biometris – Plant Research International, Wageningen, the Netherlands). Cluster analysis was run using PRIMER version 5.1.0 (PRIMER-E Ltd., Plymouth, UK). Habitat Many of the vegetation data were not normally distributed, even after transformations, so using mixed-model ANOVA with block as the random variable was not feasible. To determine whether blocks were ecologically similar based on vegetation criteria, a hierarchical cluster analysis was conducted to compare floristic and physiognomic characteristics among sites (Darveau et al. 1995). Since this analysis showed that sites were not grouped ecologically by block, block was not used as a random variable for further vegetation analyses (Mosley 2004). A PCA was conducted on 12 habitat variables summarizing structural attributes, to meet the general rule of a 3:1 ratio of samples to variables (McGarigal et al. 2000). To test the null hypothesis that vegetation structure is the same between riparian and upland habitats, independent t tests or Mann–Whitney U tests were calculated. Bird assemblages Analyses of bird data were conducted with a mixed-model ANOVA (Zar 1999). Sampling block was used as the ran-
Can. J. For. Res. Vol. 36, 2006
dom variable, because blocks were chosen from a larger population of potential forest sites (Bennington and Thayne 1994). To test the null hypothesis that riparian and upland sites were equal in species richness and bird capture rates, ANOVAs were calculated for bird species richness and an index of avian abundance during each season (spring migration, breeding, and fall migration). Each bird was assigned a migratory status (e.g., Neotropical migrants, temperate migrants, or residents) based on information from the literature (Weeber 1999; Downes and Collins 2003). Indices of bird abundance and species richness based on migratory status were compared for each habitat using χ2 tests for spring migration, breeding, and fall migration periods. To examine movement patterns, a χ2 test was calculated for each type of net (riparian parallel, riparian perpendicular, upland-A, upland-B), comparing the total index of abundance during each season. Chi-square tests on data from the total sample of birds captured were used to avoid biasing overall trends because of small samples on any given day (Townend 2002). Bird–habitat relationships A DCA was conducted on the index of abundance data from bird species captured at ≥10% (4/36) of the sites during the breeding season (16 species) to determine whether the gradient length was ≤4 standard deviations (SD), suggesting linear species response curves (ter Braak and Šmilauer 2002; Leps and Šmilauer 2003). As this was the case, an RDA was carried out, using forward selection with 999 Monte Carlo permutations (ter Braak and Šmilauer 2002; Leps and Šmilauer 2003) to identify which habitat variables contributed significantly to the index of abundance of the 16 bird species included in the analysis. An additional Monte Carlo permutation test was conducted to determine the significance of the canonical axes. Ordinations were interpreted using the biplot rule (ter Braak and Šmilauer 2002; Leps and Šmilauer 2003). Insects To test the null hypothesis that the index of insect abundance was the same between riparian and upland areas, analyses were conducted using a mixed-model ANOVA with block as the random variable. As some samples were destroyed by bears or wind, using block as a random variable was not always feasible, as few samples were left for some of the blocks. In those cases a general linear model was used.
Results Habitat The first three principal components accounted for 62.5% of the total variation of the vegetation structure variables (Fig. 3; PC1 = 29.3%, PC2 = 20.1%, PC3 = 13.2%). Sites at the positive end of the first component (PC1) had high deciduous shrub densities dominated by species normally associated with riparian habitats (e.g., leatherleaf, sweet gale, Labrador tea), open canopies, and a dense, species-rich herbaceous layer. Sites at the negative end of PC1 had closed canopies with an abundant deciduous tree component, more © 2006 NRC Canada
Mosley et al.
1153
Fig. 3. Principal components analysis of the 36 sites and the 12 summary vegetation variables near White River, Ontario, Canada (Dens, density; BA, basal area; Dec, deciduous). Sites are represented by circles (riparian) or crosses (upland). PC1 is a gradient between canopy cover on the negative end and deciduous shrub density and herbaceous cover on the positive end. PC2 is a gradient between coniferous dominance of both shrubs and trees on the negative end, and sapling density, species richness, and snag basal area on the positive end.
typical of the upland mixed forest of the region. Sites at the negative end of PC2 had a large coniferous component in the canopy and understory layers and were less species rich than sites at the positive end. Sites at the positive end of PC2 also had high sapling cover and high snag basal area. Thus, PC2 is a gradient between coniferous-dominated forest and open deciduous-dominated forest with advanced regeneration. Sites at the positive end of PC3 were species rich, had high herbaceous and low shrub cover, while sites at the negative end had a dense high shrub layer and high deciduous tree basal areas. Riparian sites were more diverse and variable in vegetation characteristics than upland sites (Fig. 3). Riparian sites were broadly classified into 14 different FEC ecosite types, whereas upland sites were classified into only 5 ecosite types (Mosley 2004). Riparian sites also had higher coefficients of variation than upland sites for 22 of the 32 species and habitat variables presented in Tables 1 and 2. Plant species richness was also higher in riparian sites (riparian:
28.5 ± 1.6; upland: 23.8 ± 0.9; t[34] = 2.523, p = 0.016), whereas canopy cover and total tree density were greater in upland sites (Table 1). Tree composition and basal areas were broadly similar between upland and riparian sites (Table 1), although trembling aspen had six times greater basal area in upland than riparian sites. This difference accounted for most of the greater basal area of all deciduous species in the upland sites (Table 1). Shrub density was higher in riparian sites, due, in part, to significantly higher densities of alder, leatherleaf, sweet gale, and wild red raspberry (Table 2). Bird assemblages A total of 862 birds of 47 species were captured in the six study blocks (Appendix A). The most abundant species was magnolia warbler (Dendroica magnolia (Wilson)) (15% of all captures), followed by Swainson’s thrush (Catharus ustulatus (Nuttall)) (10% of all captures), and yellowrumped warbler (Dendroica coronata L.) (8% of all captures © 2006 NRC Canada
1154
Can. J. For. Res. Vol. 36, 2006 Table 1. Habitat characteristics per hectare for 18 riparian and 18 upland sites near White River, Ontario, Canada, in 2003. Riparian
Upland
Habitat characteristic
Mean±SE
CV (%)
Mean±SE
CV (%)
Total tree density Canopy height (m) Basal area (m2/ha) Coniferous species Balsam fir (Abies balsamea (L.) Mill.) Black spruce (Picea mariana (Mill.) BSP) Jack pine (Pinus banksiana Lamb.) White spruce (Picea glauca (Moench) Voss) Subtotal Deciduous species Trembling aspen (Populus tremuloides Michx.) White birch (Betula papyrifera Marsh.) Subtotal Total Cover (%) Canopy (>5 m) Sapling (2–5 m)
587.5±61.7 18.3±0.5
44.6 11.4
800.0±65.6 17.5±0.7
34.8 16.8
2.5±0.7 5.6±1.1 1.2±1.0 0.8±0.5 10.2±1.3
115.4 82.3 372.6 228.0 54.4
2.2±0.5 6.0±1.5 2.3±1.4 0.0±0.0 10.5±1.7
102.0 109.7 250.8 424.8 67.5
0.7±0.4 4.0±1.2 4.7±1.3 19.0±1.9
271.0 125.3 114.6 43.4
4.5±2.1 5.6±1.0 10.3±2.3 24.5±1.9
34.2±3.1 34.2±5.5
38.3 68.1
46.1±2.7 31.9±4.0
t[34]
p
–2.356
0.024 0.350*
–0.152
0.962* 0.681* 0.903* 0.127* 0.880
202.0 74.1 94.6 32.7
–2.035
0.023* 0.114* 0.025* 0.050
24.8 53.2
–0.510
0.018* 0.960†
Note: Calculations were made using independent t tests except where otherwise noted. Significant p values are shown in bold. Red maple (Acer rubrum L.) and eastern larch (Larix laricina (Du Roi) K. Koch) were found in small proportions on some sites and are included in appropriate totals. *Calculations made using Mann–Whitney U test. † Data were log10 transformed to meet the assumptions of t test.
Table 2. Herbaceous cover and shrub density and cover of 18 riparian and 18 upland sites near White River, Ontario, Canada, in 2003. Riparian Mean±SE Shrub density (stems/ha) Coniferous species Balsam fir (Abies balsamea (L.) Mill.) Black spruce (Picea marina (Mill.) BSP) Subtotal Deciduous species Alder (Alnus spp.)* Beaked hazel (Corylus cornuta Marsh.) Bush honeysuckle (Diervilla lonicera Mill.) Labrador tea (Ledum groenlandicum Oeder) Leatherleaf (Chamaedaphne calyculata (L.) Moench) Mountain ash (Sorbus americana Marsh.) Mountain maple (Acer spicatum Lam.) Serviceberry (Amelanchier spp.) Sweet gale (Myrica gale L.) White birch (Betula papyrifera Marsh.) Wild red raspberry (Rubus idaeus L. var. strigosus (Michx.) Maxim.) Subtotal Total Cover (%) High shrub (0.5–2 m) Low shrub (0–0.5 m) Herbaceous
Upland CV (%)
Mean±SE
CV (%)
t[34]
p
526.7±107.8 208.9±67.3 740.0±111.2
86.8 136.8 63.8
668.9±141.7 113.3±42.4 822.2±161.8
89.9 158.9 83.5
0.475 0.309 0.987
1093.3±398.3 257.8±153.2 306.7±152.2 482.2±237.8 1468.9±728.5 131.1±34.5 313.3±143.6 308.9±102.9 1006.7±433.2 375.6±199.8 286.7±129.8
154.5 252.1 210.6 209.2 210.4 111.6 194.4 141.4 182.6 225.7 192.1
131.1±100.8 415.6±231.9 426.7±192.1 104.4±62.4 0±0 164.4±38.0 528.9±183.1 324.4±98.4 0±0 128.9±57.8 0±0
326.0 236.8 209.2 253.4 0 98.0 146.9 126.7 0 190.3 0
0.006 0.920 0.427 0.212 0.018 0.491 0.874 0.886 0.018 0.565 0.001
6644.4±1344.2 7384.4±1330.3
85.8 76.4
3728.9±479.1 3346.7±466.7
80.5 59.1
0.005 0.012
51.8±4.8 56.4±3.8 59.1±6.0
39.7 28.8 42.8
51.7±3.8 52.8±3.8 47.5±4.0
31.1 30.5 35.5
0.018 0.670
0.986 0.508 0.194
Note: Calculations were made using Mann–Whitney U tests except where otherwise noted. Significant p values are shown in bold. *Green alder (Alnus crispa (Ait.) Pursh) and speckled alder (Alnus rugosa (Du Roi) Spreng.).
© 2006 NRC Canada
Mosley et al.
1155
Table 3. Species richness and the index of avian abundance in 18 riparian and 18 upland sites near White River, Ontario, Canada, in 2003. ANOVA summary (α = 0.1) Frequency* Riparian
Habitat Upland
F[1,5]
Spring migration Index of avian abundance Species richness
6.9±1.0 (0–13) 4.4±0.6 (0–9)
6.9±1.4 (0–22) 3.8±0.7 (0–10)
Breeding season Index of avian abundance Adults Juveniles Species richness Adults Juveniles
9.2±1.2 6.1±0.9 2.3±0.6 5.9±0.6 4.3±0.5 1.9±0.4
6.8±1.4 4.3±0.7 2.1±0.7 3.9±0.7 2.9±0.5 1.5±0.5
Fall migration Index of avian abundance Species richness
(3–19) (1–17) (0–10) (2–11) (1–8) (0–6)
11.3±4.5 (0–83) 5.1±1.3 (0–25)
(0–22) (0–9) (0–13) (0–11) (0–6) (0–8)
6.9±1.5 (0–22) 4.4±0.8 (0–12)
Block
Interaction
p
F[5,5]
p
F[5,24]
p
0.233 0.789
0.650 0.415
0.578 1.114
0.719 0.454
0.883 0.676
0.507 0.646
6.471 5.042
0.052 0.075 0.571† 0.004 0.004 0.379†
6.193 2.612
0.033 0.158
0.292 0.416
0.913 0.833
9.683 7.937
0.013 0.020
0.229 0.167
0.946 0.972
0.561 0.796
1.946 1.298
0.241 0.391
1.258 1.083
0.314 0.395
24.359 26.406
0.387 0.075
Note: Mixed-model ANOVAs using block as the random variable were used in all calculations, as there was a significant block effect during the breeding period. Where assumptions of ANOVA were not met, Mann–Whitney U tests were used. Significant p values are shown in bold. *Frequency is the number of species per site for species richness and the number of captures per site for the index of avian abundance. Values are means ± SEs with ranges given in parentheses. † Mann–Whitney U tests.
(Appendix A). Three species, black-backed woodpecker (Picoides arcticus (Swainson)), northern parula (Parula americana L.), and white crowned sparrow (Zonotrichia leucophrys (Forster)), were captured only once during the sampling period (Appendix A). The index of avian abundance, as measured by average captures per site, was not statistically significantly different between riparian and upland sites during neither spring nor fall migration (Table 3). During the breeding period, the index of avian abundance was higher in riparian than in upland sites (F[1,5] = 6.471, p = 0.052), primarily due to a higher abundance of adult birds (Table 3). Similar patterns were found for species richness, with no significant differences found in either spring (F[1,5] = 0.789, p = 0.415) or fall (F[1,5] = 0.075, p = 0.796) migration periods. However, during the breeding period, species richness was significantly greater in riparian than in upland sites (F[1,5] = 24.359, p = 0.004), due to greater species diversity of adult birds (Table 3), and possibly as a consequence of the larger total number of birds sampled. Species found exclusively in riparian areas included common yellowthroat (Geothlypis trichas L.), northern waterthrush (Seiurus noveboracensis (Gmelin)), and swamp sparrow (Melospiza georgiana (Latham)) (Appendix A). Species that were found mostly (>70% of all captures) in riparian habitats included American redstart (Setophaga ruticilla L.), Canada warbler (Wilsonia canadensis L.), chestnut-sided warbler (Dendroica pensylvanica L.), gray jay (Perisoreus canadensis L.), red-eyed vireo (Vireo olivaceus L.), Tennessee warbler (Vermivora peregrina (Wilson)), winter wren (Troglodytes troglodytes L.), and yellow-bellied sapsucker (Sphyrapicus varius L.) (Appendix A). Conversely, American robin (Turdus migratorius L.), black-backed woodpecker (Picoides arcticus Swainson), and least flycatcher (Empidonax mini-
mus (Baird and Baird)) were found exclusively in upland habitats, while blackpoll warbler (Dendroica striata (Forster)), black-throated blue warbler (Dendroica caerulescens (Gmelin)), and ovenbird (Seiurus aurocapillus L.) were found mostly (>70% of all captures) in upland habitats (Appendix A). All three migratory guilds (Neotropical migrants, temperate migrants, and permanent residents) were found in each habitat, although during the fall migration period Neotropical migrants made up a significantly larger proportion of the 2 bird community in riparian than in upland sites (χ[0.05] = 8.8, df = 2, p = 0.012) (Fig. 4). Combining all time periods, species richness was not significantly different, for any migratory guild, between riparian and upland habitat. During spring migration, capture rates were similar for all 2 = 0.333, df = 3, p = 0.954), although four net types (χ[0.05] during the breeding and fall migration periods, more birds were caught in the riparian nets set perpendicular to the 2 streams than in the other three net types (breeding: χ[0.05] = 2 16.7, df = 3, p < 0.001; fall: χ[0.05] = 34.2, df = 3, p < 0.001) (Fig. 5). Bird–habitat relationships The first four axes generated by the RDA explained 29.8%, 17.4%, 13.6%, and 10.4% of the variability in the species–environment relationship, respectively. All axes were statistically significant (p = 0.001), although only the first two were interpreted biologically. The RDA biplots (Fig. 6) identified a distinct gradient between tree and shrub densities along the first axis, with the positive end having high proportions of shrubs (including serviceberry and raspberry), and the negative end having high poplar and total tree density. Both upland and riparian sites were found in high proportions on the negative side of the first axis, although the positive side contained mostly © 2006 NRC Canada
1156
Can. J. For. Res. Vol. 36, 2006
Fig. 4. Illustration of index of avian abundance organized by migratory status during migration and breeding periods near White River, Ontario, Canada.
Fig. 5. Comparison of total number of birds captured in different net types near White River, Ontario, Canada (Rip-Perp, riparian net perpendicular to the stream; Rip-Par, riparian net parallel to the stream; Up-A, first upland net; Up-B, second upland net).
riparian sites (seven riparian compared with four upland) (Fig. 6a). The gradient on the second axis is not as distinct, but sites with high densities of mountain maple tend to fall on the positive end, and those with high beaked hazel densities tend to fall on the negative end. Ten vegetation variables, including mountain maple density, sapling cover, serviceberry density, total snag basal area, deciduous shrub density, raspberry density, beaked ha-
zel density, snag density (in decay class 1), poplar density, and total tree density, were identified as significantly contributing to the index of abundance of the 16 bird species in the RDA. Magnolia warbler, Canada warbler, Nashville warbler (Vermivora ruficapilla (Wilson)), black-throated blue warbler, American redstart, and Swainson’s thrush were positively correlated with sapling cover, serviceberry density, deciduous shrub density, and raspberry density, whereas © 2006 NRC Canada
Mosley et al.
1157
Fig. 6. Ordination biplots of (a) sites (riparian, circles; upland, crosses), and (b) bird species captured during the breeding period (24 June – 15 Aug.) near White River, Ontario, Canada. Vegetation variables were identified as significantly contributing to the index of abundance of 16 bird species in the redundancy analysis (RDA).
© 2006 NRC Canada
1158
Can. J. For. Res. Vol. 36, 2006 Table 4. Total index of insect biomass captured by sticky traps in riparian and upland habitats. Index of biomass (mg/site)* Time period
Factor
F
df
p
Riparian
Upland
(1) 26 May – 16 June
Habitat Block Interaction Habitat Block Interaction Habitat Habitat Habitat Habitat
