Moving riparian management guidelines towards a natural disturbance model: An example using boreal riparian and shoreline forest bird communities

Forest Ecology and Management 257 (2009) 54–65

Contents lists available at ScienceDirect

Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco

Moving riparian management guidelines towards a natural disturbance model: An example using boreal riparian and shoreline forest bird communities Kevin J. Kardynal a, Keith A. Hobson a,b,*, Steven L. Van Wilgenburg c, Julienne L. Morissette d a

Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, SK S7N 5E2, Canada Environment Canada, 11 Innovation Place, Saskatoon, SK S7N 3H5, Canada c Environment Canada, 115 Perimeter Road, Saskatoon, SK S7N 0X4, Canada d Ducks Unlimited Canada, #100 17958 106th Avenue, Edmonton, AB T5S 1V4, Canada b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 20 February 2008 Received in revised form 14 August 2008 Accepted 14 August 2008

Forest harvesting strategies that approximate natural disturbances have been proposed as a means of maintaining natural species’ diversity and richness in the boreal forests of North America. Natural disturbances impact shoreline forests and upland areas at similar rates. However, shoreline forests are generally protected from harvest through the retention of treed buffer strips. We examined bird community responses to forest management guidelines intended to approximate shoreline forest fires by comparing bird community structure in early (1–4 years) post-burned and harvested boreal riparian habitats and the adjacent shoreline forest. We sampled riparian areas with adjacent: (1) burned merchantable shoreline forest (n = 21), (2) burned non-merchantable shoreline forest (n = 29), (3) 10 m treed buffer with 25% retention in the next 30 m (n = 18), and (4) 30 m treed buffer (n = 21). Only minor differences were detected in riparian species’ abundance and bird community composition between treatments with greater differences in these parameters occurring between post-fire and post-harvest upland bird communities. Indicators of all merchantable treatments were dominated by upland species with open-habitat species and habitat generalists being typical upland indicator species of burned merchantable habitats and forest specialists typical upland indicators of harvested treatments. Riparian species indicative of burned riparian habitats were Common Yellowthroat (Geothlypis trichas), Le Conte’s Sparrow (Ammodramus leconteii) and Eastern Kingbird (Tyrannus tyrannus) and indicators of 30 m buffers were Alder Flycatcher (Empidonax alnorum) and Wilson’s Warbler (Wilsonia pusilla). Multivariate Redundancy Analysis (RDA) of the overall (riparian and upland birds) community showed greater divergence than RDA with only riparian species suggesting less effect of fire and forestry on riparian birds than on upland birds. Higher natural range of variability (NRV) of overall post-fire bird communities compared to post-harvest communities emphasizes that harvesting guidelines currently do not achieve this level of variability. However, lack of a large negative effect on common riparian species in the first 4 years post-disturbance allows for the exploration of alternative shoreline forest management that better incorporates bird community composition of post-fire riparian areas and shoreline forests. ß 2008 Elsevier B.V. All rights reserved.

Keywords: Bird community Boreal plain Fire Forestry Natural disturbance hypothesis Natural range of variation Riparian

1. Introduction Increasing anthropogenic pressures in the North American boreal forest (Cumming et al., 1994; Timoney, 2003) have necessitated the adoption of forest management practices that attempt to maintain natural ecosystem processes (Hunter, 1993; Attiwill, 1994). Recent research has highlighted the impacts of such anthropogenic alterations on boreal avifauna (e.g., Welsh,

* Corresponding author at: Environment Canada, 11 Innovation Place, Saskatoon, SK S7N 3H5, Canada. Tel.: +1 306 975 4102; fax: +1 306 975 5143. E-mail address: [email protected] (K.A. Hobson). 0378-1127/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2008.08.029

1987; Kirk et al., 1997; Hobson and Bayne, 2000). Over 200 species of birds nest in the western Canadian boreal forest (Smith, 1993), many of these show their highest abundance there (Blancher, 2003) and few are considered threatened (Dunn, 2002). As such, conservation practices that focus on singles-species management are neither practical nor feasible, and a coarse-filter approach (e.g., one that considers multiple species) to forest management is more appropriate (Armstrong et al., 2003). Current hypotheses proposed to maintain wildlife populations in managed forests suggest using natural disturbance regimes to guide forest harvesting operations (Hunter, 1993). This natural disturbance hypothesis assumes that forest biota will be more likely to adapt to anthropogenic disturbances that retain structural

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

attributes similar to the natural disturbances with which the species have presumably evolved (Attiwill, 1994). Fire is considered the dominant disturbance agent in the boreal forest (Rowe and Scotter, 1973) and attempts to mimic disturbances using forestry have primarily used wildfire as a model (DeLong, 2002). In efforts to align forest management with ecologically based paradigms, modifications to conventional forestry practices have included cutting larger areas (aggregating cutblocks), retaining live residual trees within cutblocks and, more recently, cutting closer to water bodies (DeLong and Tanner, 1996; Saskatchewan Environment, 2006b). Riparian areas are considered the most productive and species rich environments on many landscapes (Naiman et al., 1993; Sabo et al., 2005) and in the boreal forest are used by over 40% of the bird species for nesting and foraging (K.J. Kardynal, unpublished data). These areas are subject to frequent and recurring disturbances including flooding, herbivory by beavers and other large mammals and fire (Naiman et al., 1993; Macdonald et al., 2004; Martell et al., 2006). On the presumption that they have a disproportionately high conservation value (Gregory et al., 1991; Decamps et al., 2004), riparian areas have generally been excluded from efforts to approximate patterns of natural disturbances (i.e., harvesting or burning shoreline forest). The most common approach in boreal forest harvesting is to leave strips of trees (buffers) adjacent to water bodies with the primary objective of protecting aquatic environments (Hickey and Doran, 2004). Thus, conventional management of shoreline forests using buffer retention is not consistent with efforts

55

to implement harvesting practices that approximate natural disturbances because such disturbances occur in shoreline forests and riparian areas at rates similar to upland forest (Harper and Macdonald, 2001; Macdonald et al., 2004). North American studies of bird responses to disturbance in riparian areas have centered on the ability of buffers of different widths to provide habitat for upland-nesting birds (e.g., Darveau et al., 1995; Hannon et al., 2002). Additionally, tests of the natural disturbance hypothesis using avifauna have focused on bird communities in upland habitats (reviewed in Schieck and Song, 2006). These studies have shown high disparity in avian community composition between early (1–5 years) post-fire and post-harvest landscapes as a result of dissimilarities in habitat structure. Convergence of community composition between these disturbance types generally occurs 30–60 years post-disturbance (Hobson and Schieck, 1999; Imbeau et al., 1999; Simon et al., 2002). No previous studies have contrasted bird communities inhabiting early post-fire and post-harvest shoreline forest habitats and so it is not clear how current riparian management guidelines depart from the Natural disturbance hypothesis for boreal forest riparian area and shoreline forest birds. The objectives of this study were to determine if new forest management guidelines in Saskatchewan intended to more closely approximate fires in shoreline forests maintain riparian and upland bird communities similar to early post-fire habitats. Specifically, we examined bird species’ abundance and frequency of occurrence and community composition in riparian areas with adjacent: (1) burned merchantable shoreline

Fig. 1. Locations of study sites within the Boreal Plain ecozone (shaded area) of western Canada. Harvested sites were interspersed between sampled fires in Saskatchewan. Year of fire indicated in parentheses.

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

56

2. Methods and study design

riparian-associated species would respond to riparian vegetation parameters (e.g., structure, species composition, riparian width) rather than wetland size. Therefore, 12 larger lakes (600 ha) that exhibited all of the above criteria were included to increase the sample size.

2.1. Study design

2.2. Avian surveys

Study sites (4 fires and 20 cutblocks) were distributed across the Boreal Plain ecozone of western Canada (Acton et al., 1998; Fig. 1) from the House River Fire, Alberta (558760 N 1128140 W) to Candle Lake, Saskatchewan (548730 N 1038 690 W). The Boreal Plain is a mosaic of various upland forest stand ages and types including hardwoods (Trembling Aspen Populus tremuloides, Balsam Poplar Populus balsamifera, White Birch Betula papyrifera), softwoods (White Spruce Picea glauca, Jack Pine Pinus banksiana, Balsam Fir Abies balsamea) and mixedwoods. Wetlands cover 30–50% of the Boreal Plain and include marshes, swamps, fens and bogs (National Wetlands Working Group, 1997). Open water wetlands were selected in four treatments based on merchantability and age (>80 years) of the adjacent shoreline forest, proportion of the shoreline forest harvested, and based on whether adjacent riparian zone and shoreline forests were disturbed by fire or harvesting. Wetlands with adjacent merchantable shoreline forest contained >70% harvestable shoreline timber (trees >15 cm diameter at breast height, >15 m tall, with dry soils) prior to disturbance (assessed via forest inventory data and ground-truthed where possible) within 50 m of the survey transect line. Sites with burned non-merchantable shoreline forest were used to jointly assess, with burned merchantable sites, the natural range of variability (NRV, i.e., abundance and community composition) of bird communities in early post-fire riparian habitats. Burned merchantable sites were used to contrast different buffer widths with natural disturbance. Harvested sites were located in two Forest Management Areas (FMAs) in Saskatchewan with different riparian management guidelines, 10 m buffer with 25% tree retention in the adjacent 30 m (hereafter, 10 m buffer) and 30 m buffer (Saskatchewan Environment, 2006a,b; Table 1) and were interspersed between the sampled fires (Fig. 1). All samples in the burned merchantable treatment were at least 60% burned (including shrub and forested areas) to match the amount of disturbance in the harvested treatments. The majority of sites on the landscape that met sampling criteria (below) were surveyed and so sites were not selected randomly. All selected riparian areas and adjacent shoreline forests were surveyed within 4 years of disturbance, had disturbance (fire or harvest) within 50 m of the wetland’s high water mark, included disturbance along at least 400 m of the wetland, and possessed a riparian zone (i.e., riparian vegetation) 5–15 m wide (50 m on wetlands with non-merchantable shoreline forest). Target wetland size was 2–30 ha; however, we assumed that wetland-obligate and

Birds were surveyed using a combined variable-width strip transect/variable-radius point-count technique (Hobson and Schieck, 1999). Transects 400 m in length were placed parallel to the upland-riparian ecotone along the disturbed portion (burned or harvested) of the shoreline forest. Two point-counts were placed at the 50 and 350 m mark of the transect, 50 m from the water’s edge (Fig. 2). Ten-minute point-counts were divided into three time intervals (0–3, 3–5, 5–10 min) for detection probability estimation using the count-removal method (Farnsworth et al., 2002). Distance to birds was estimated on transects and point-counts in bands of 0–25, 26–50, 51–100 or >100 m from the observer. Individual birds detected on both the point-count and transect were identified during the survey and counted as only one bird in all analyses. Each site was surveyed once and all treatments were interspersed throughout the sampling period in both years of the study to eliminate the effects of seasonality of bird activity (i.e., changes in singing rate). Observers were trained together prior to sampling and were alternated between treatments to avoid confounding treatment and observer effects. Surveys started at sunrise and ceased 5 h after sunrise during 1–30 June 2004 and 2005 and were performed only on days with fair weather (no rain and winds <4 on a Beaufort scale). Survey start times were varied among treatments to prevent biases in species’ detection.

forest (n = 21), (2) burned non-merchantable shoreline forest (n = 29), (3) 10 m treed buffer with 25% retention in the next 30 m (n = 18), and (4) 30 m treed buffer (n = 21).

2.3. Vegetation surveys Generalized features of vegetation composition and structure were collected at each sampling area at 100 m intervals along each survey transect (four survey stations per transect) and encompassed an area from the open water to 50 m into the upland (Table 2). Measurements included: (i) canopy composition (minimum of 20% of canopy; hardwood, softwood, hardwooddominated mixedwood, softwood-dominated mixedwood, fen, bog) determined visually; (ii) percent canopy closure assessed visually; (iii) canopy height by calibrating using a clinometer; (iv) dominant shrub species (five most abundant species; minimum of 20% of community); (v) average shrub height within each plot in four height classes (0–0.5, 0.5–1, 1–3 and 3–10 m); (vi) rank ground cover of herb, moss and grass in the upland; and (vii) grass, sedge and riparian vegetation width. Number of snags >2 m in height and >10 cm diameter at breast height (DBH) was recorded and classified into decay stages according to Lee et al. (1997). Downed woody material (DWM) >10 cm DBH was measured at

Table 1 Description of treatment classes used in the study, their sample sizes (N) and mean (2S.D.) wetland size (ha) from the Boreal Plain ecozone of Saskatchewan and Alberta, 2004 and 2005 Treatment a

Burned merchantable Burned non-merchantable 10 m bufferb 30 m buffer

N

Descriptiona

21 29 18 21

>70% <70% >50% <50%

burned merchantable forest burned non-merchantable forest merchantable forest harvested merchantable forest harvested

Mean (2.S.D) wetland size (ha) 47.2 11.7 86.4 27.6

(100.8) (12.8) (193.2) (112.6)

a Merchantability for burned treatments was defined as the proportion of a 50 m digital buffer around the transect survey line considered merchantable pre-disturbance (>70% of the trees, >15 cm diameter at breast height, >15 m tall, forest composition consisting of >70% hardwood or softwood species with dry soils) from digital forest inventory data. Percent harvested was calculated as the proportion of the merchantable timber harvested relative to pre-disturbance. b 10 m buffer with 25% retention in the next 30 m.

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

57

Fig. 2. Transect (400 m in length) and point-count station placement adjacent to riparian areas along wetlands for bird surveys performed in the Boreal Plain of Saskatchewan and Alberta. Transects were placed at the riparian/upland forest edge and point counts were placed at 50 and 350 m along the transect, 50 m from the water’s edge.

two points (centered at the point-count stations) 50 m from the water’s edge and transect end using a line intercept method in two random directions. On burned sites, burn severity measures were collected on the canopy and ground (see Table 2). A wetland classification was assigned to each site (the length of the transect) expressed as percentages of each wetland type according to Harris et al. (1996). Amount of burned graminoid vegetation was not collected because accurate measurements could not be taken due to re-growth of this vegetation type after fire. Data for each measurement were averaged for the four survey stations for use in data analysis. Vegetation sampling was conducted from 15 June to 30 July 2004 and 15 June to 30 August 2005. To increase the resolution of digital forest inventories of riparian areas, vegetation features (survey line, cutblock boundary, live treed residual patches and shrub line-shrub/

riparian and shrub/tree line boundaries) were mapped at each wetland using Geographic Positioning Systems (GPS) units. Based on GIS analysis of digital forest cover maps, we calculated percent retention as the proportion of the merchantable forest not disturbed (harvested or burned) in a 50 m rectangular GIS buffer around the transect survey line. 2.4. Data analysis 2.4.1. Wetland size and vegetation We tested for differences in wetland size using analysis of variance (ANOVA) with Bonferonni post hoc tests to determine between-treatment differences. Principle components analysis (PCA) was used to explore the inter-correlation between vegetation variables and to determine whether the number of variables

Table 2 Vegetation parameters measured along the riparian/upland interface at sampling areas early post-harvest and/or post-fire in the Boreal Plain ecozone of Alberta and Saskatchewan, 2004 and 2005 Vegetation parameter

Description

Stand type Canopy Canopy height Closure Sub-canopy Snags Shrub species Shrub cover Shrub height Ground cover DWM Riparian Emergent vegetation Canopy burn severity

Dominant (>70% of stand) forest type (hardwood, softwood, mixedwood, fen, bog, other) Dominant tree species [minimum 20% of canopy (>15 m in height)] Average height of canopy % Canopy closure Five most abundant species [minimum 20% of community (10–15 m in ht.)] % Stand dead; decay stage of majority of dead trees Five most abundance species (minimum 20% of community) % Cover in: (1) 0–0.5 m; (2) 0.5–1.0 m; (3) 1–3 m; (4) 3–10 m height classes Average height of all shrubs Rank dominant cover (1–3) of herb, moss and grass Downed woody material—no. of downed logs >10 cm diameter intercepted in two 50 m line transects Average riparian width, height and dominant species Average emergent vegetation width, height and dominant species Burn severity category on features of the canopy: 0, unburned; 1, >60% of trees with green needles/leaves; 2, 40–60% of trees with green needles/leaves; 3, 5–40% of trees with green needles/leaves; 4, <5% of trees with green needles/leaves; 5, all trees having brown (dead) or no needles/leaves; 5, all trees having brown (dead) or no needles/leaves; 6, mostly broken stumps with a few standing dead trees Burn severity category on features of the ground: 0, no burn evidence on the forest floor; 1, light or limited charring of duff/moss; 2, substantially charred duff/moss; 3, extensive exposure of mineral soil Average height of charring on tree boles % wetland classification type of each wetland Swamp (shrubby, hardwood, softwood, mixedwood), marsh, bog (shrubby, treed), fen (graminoid, shrubby, treed), etc.

Ground burn severity Burn height Wetland class

Data was collected at 100 m intervals along each 400 m survey transect and encompassed an area from the open water, 50 m into the upland.

58

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

could be reduced to fewer explanatory components. Separate PCA analyses were run for pre- and post-disturbance GIS data, field observations, and wetland classification data. Pre- and postdisturbance proportion data were arcsine transformed prior to analysis to improve non-normally distributed data. The retained principle components were selected using the broken-stick method (McCune and Mefford, 1999). All multivariate data were analyzed with PC-ORD v.4.0 (McCune and Mefford, 1999) except where noted. PCA scores were compared for between-treatment differences using a simple one-way analysis of variance (ANOVA) using Bonferonni post hoc tests in SPSS for Windows, Release 14.0.0. (2005) with a level of p < 0.01 used to indicate significant differences. 2.4.2. Avian Point-count data were adjusted for biases in detection probability using Farnsworth et al. (2002) count-removal method for 11 species (>80 observations). For 83 species with low sample sizes (<80 detections), species were classified into 27 groups based on song characteristics, behavioural traits and habitat preferences (Alldredge et al., 2007; Table 3). Detectability models were run using Huggins (1989) closed-capture models in program R (R Core Development Team, 2007) with RMark v.1.6.4 (Laake, 2007), an interface to program MARK (White, 2006). Detectability was modeled as a linear function including date, time of day, observer ability (relative ability to identify species), distance to bird, percent of open habitat as covariates along with interaction terms. Models were selected based on AICc values (Akaike’s Information Criteria adjusted for small sample size) and selected if DAICc was less than four (Burnham and Anderson, 2002). Where model uncertainty occurred, we used model averaging to generate parameter estimates. Details on models and detectability estimates are available from the senior author upon request. Detectability estimates were used to adjust raw counts prior to further analysis. Individuals detected flying over, waterfowl, raptors, species detected fewer than three times and outliers determined using outlier analysis in PC-ORD v.4.0 (McCune and Mefford, 1999) were excluded from all analyses. Detection probabilities were not estimated for individual birds encountered only on transect surveys due to the layout of the transects, which violated assumptions of DISTANCE sampling (Buckland et al., 2001). All birds detected on point-counts adjusted for detection probability and unadjusted transects at one sample unit were then combined for further analysis. As indicated above, we attempted to indicate when individual birds were detected by both point-count and transect surveys, and in these instances, only adjusted counts from the point-count were used, thus resulting in a single observation from both methodologies. We compared distributions of individual species across treatments and disturbance types using Indicator Species Analysis (ISA; Dufrene and Legendre, 1997). This method combines individual species’ abundances and frequency of occurrence to derive indicator values for each species in each treatment. We used a significance value of p < 0.1 for ISA because we expected some riparian species would show a relatively weak response to treatment as riparian habitat (e.g., emergent, graminoid vegetation) was generally not as affected by treatment than upland habitats (e.g., forest). To examine bird community response to disturbance types and post-disturbance vegetation variables, Redundancy Analysis (RDA) was used (ter Braak, 1986). Treatment types were entered in the vegetation matrix as dummy variables, significant pre-disturbance vegetation variables (measured from GIS data) were entered as covariates and post-disturbance vegetation components were included as continuous explanatory variables. To account for a large spatial gradient, we included the longitude from the centroid of

each wetland as a covariate in the ordination (Legendre, 1993). We created two ordinations by running RDA for all (riparian and upland) species and again, separately, for only the riparian species in CANOCO v.4.54 (ter Braak and Smilauer, 1999). Species abundance data were square-root transformed prior to analysis. Assessment of the variance inflation factors in the correlation matrix was used to assess collinearity of environmental variables (Quinn and Keough, 2004). The ordinations were symmetrically scaled and centered by species. To aid in the interpretation of the ordinations and to assess differences in community composition between treatments, 67% Confidence Ellipses (representing  1S.D.) around treatment groups were used (Hobson and Schieck, 1999). Multi-Response Permutation Procedures (MRPP) with the Euclidean Distance measure, was used to determine if differences in community composition between treatments were significant (McCune and Grace, 2002). 3. Results 3.1. Wetland size and pre-disturbance vegetation Differences in wetland size among treatments were nonsignificant (F3,88 = 1.84, p > 0.1) implying that water body size should not bias comparisons of bird community data between treatments. From the pre-disturbance vegetation data, four axes encompassed 81.9% of the cumulative variance. As indicated by their scores, Axis 1 represented amount of non-merchantable softwood, Axis 2 represented amount of graminoid vegetation and Axis 3 represented amount of hardwood forest. Comparisons of pre-disturbance PCA scores using ANOVA of these four axes showed that area of low merchantable softwood (Axis 1; F3,88 = 5.06, p < 0.01) and graminoid vegetation (Axis 2; F3,88 = 8.64, p < 0.001) differed significantly between treatments. Bonferroni post hoc tests indicated differences for low merchantable softwood between burned non-merchantable sites and 30 m buffers (p < 0.005) and for graminoid vegetation between burned non-merchantable treatments and all other treatments (p < 0.01). PCA scores for Axis 1 and Axis 2 were saved for use as covariates in analysis of the bird community data, to attempt to control for differences in pre-disturbance vegetation between treatments. PCA scores for hardwood vegetation (Axis 3; F3,88 = 4.48, p < 0.05) were included to assess the influence of hardwood vegetation on the ordinations. 3.2. Post-disturbance vegetation Analysis of post-disturbance vegetation variables derived from GIS showed the first three ordination axes sufficiently described variation in vegetation parameters representing 77.6% of the variance. Axis 1 represented amount of area disturbed, Axis 2 represented amount of graminoid vegetation and Axis 3 represented area of fen. Analysis of field data indicated that three axes explained 57.6% of overall variance: Axis 1 was associated with canopy closure, Axis 2 with upland grass cover, and Axis 3 with canopy height. PCA of the wetland classification data showed two axes (Axis 1—amount of meadow marsh and Axis 2—amount of treed fen) represented 55.0% of the cumulative variance. Tests for differences between PCA scores among treatments showed significant differences for six variables. For GIS data, area disturbed (Axis 1; F3,88 = 6.97, p < 0.001) and area of fen (Axis 3; F3,88 = 8.99, p < 0.001) showed significant between-treatment differences. Bonferonni post hoc tests indicated differences between burned merchantable sites and 30 m buffers (p < 0.01) and between non-merchantable and 30 m buffers (p < 0.001) for area disturbed and between burned non-merchantable and the two harvest treatments for area of fen (p < 0.01). From field data,

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

59

Table 3 Scientific name, four letter identification code, average detection probability estimate (DPE  1S.E.) and nesting guild for species or groups of species encountered during 10-min point-counts around small wetlands in the Boreal Plain of Saskatchewan and Alberta, 2004–2005 Species

Scientific nomenclature

Code

Average

Nesting

DPE (%)  1S.E.

Guild

Alder Flycatchera American Goldfinch American Redstart American Robin Black-and-white Warbler Black-backed Woodpecker Brown-headed Cowbird Black Terna Black-throated Green Warbler Blue-headed Vireo Boreal Chickadee Brown Creeper Cape May Warbler Clay-colored Sparrow Chipping Sparrow Connecticut Warbler Common Raven Common Snipea Common Yellowthroata Chestnut-sided Warbler Dark-eyed Junco Eastern Kingbirda Gray Jay Greater Yellowlegsa Hairy Woodpecker Hermit Thrush House Wren Killdeera Le Conte’s Sparrowa Least Flycatcher Lincoln’s Sparrowa Magnolia Warbler Mourning Warbler Nashville Warbler Northern Flicker Northern Waterthrusha Olive-sided Flycatchera Ovenbird Palm Warblera Philadelphia Vireo Pine Siskin Ruby-crowned Kinglet Red-breasted Nuthatch Red-eyed Vireo Red-winged Blackbirda Rose-breasted Grosbeak Sandhill Cranea Savannah Sparrow Soraa Solitary Sandpipera Song Sparrowa Spotted Sandpiper Swainson’s Thrush Swamp Sparrowa Tennessee Warbler Tree Swallowa Warbling Vireo Western Wood-Pewee Wilson’s Warblera Winter Wren White-throated Sparrow Yellow-bellied Sapsucker Yellow-headed Blackbirda Yellow Warblera

Empidonax alnorum Carduelis tristis Setophaga ruticilla Turdus migratorius Mniotilta varia Picoides arcticus Molothrus ater Chlidonias niger Dendroica virens Vireo solitarius Poecile hudsonicus Certhia Americana Dendroica tigrina Spizella pallida Spizella passerina Oporornis agilis Corvus corax Gallinago gallinago Geothlypis trichas Dendroica pensylvanica Junco hyemalis Tyrannus tyrannus Perisoreus canadensis Tringa melanoleuca Picoides villosus Catharus guttatus Troglodytes aedon Charadrius vociferus Ammodramus leconteii Empidonax minimus Melospiza lincolnii Dendroica magnolia Oporornis philadelphia Vermivora ruficapilla Colaptes auratus Seiurus noveboracensis Contopus cooperi Seiurus aurocapillus Dendroica palmarum Vireo philadelphicus Carduelis pinus Regulus calendula Sitta canadensis Vireo olivaceus Agelaius phoeniceus Pheucticus ludovicianus Grus canadensis Passerculus sandwichensis Porzana carolina Tringa solitaria Melospiza melodia Actitis macularius Catharus ustulatus Melospiza georgiana Vermivora peregrina Tachycineta bicolor Vireo gilvus Contopus sordidulus Wilsonia pusilla Troglodytes troglodytes Zonotrichia albicollis Sphyrapicus varius Xanthocephalus xanthocephalus Dendroica petechia

ALFL AMGO AMRE AMRO BAWW BBWO BHCO BLTE BTNW BHVI BOCH BRCR CMWA CCSP CHSP CONW CORA COSN COYE CSWA DEJU EAKI GRAJ GRYE HAWO HETH HOWR KILL LCSP LEFL LISP MAWA MOWA NAWA NOFL NOWA OSFL OVEN YBHL PHVI PISI RCKI RBNU REVI RWBL RBGR SACR SAVS SORA SOSA SOSP SPSA SWTH SWSP SWTH TRES WAVI WEWP WIWA WIWR WTSP YBSA YHBL YWAR

81.94  0.30 76.15  0.79 78.46  0.85 79.31  0.79 81.21  0.34 77.85  0.50 75.18  1.8 n/a 69.35  0.33 85.27  0.08 61.74  1.59 73.55  1.19 79.81  0.48 80.60  0.15 77.87  0.05 84.84  0.18 75.60  0.27 82.68  1.15 91.57  0.31 78.92  0.77 77.87  0.05 67.44  1.43 69.94  0.51 76.34  0.35 77.85  0.50 80.12  0.43 79.83  0.30 76.34  0.30 81.21  0.34 88.78  0.18 83.40  0.89 80.29  0.55 84.84  0.18 86.67  0.62 77.85  0.50 75.41  0.11 81.94  0.30 88.56  0.49 79.04  0.47 85.27  0.08 n/a 61.74  0.57 61.74  1.64 70.76  0.08 79.04  0.66 68.23  0.61 n/a 81.21  0.34 81.53  0.56 76.34  0.35 77.45  0.35 76.34  0.35 80.12  0.43 80.60  0.15 80.12  0.62 86.67  1.43 67.44  0.08 91.06  0.33 80.29  0.55 79.83  0.30 79.81  0.17 79.43  0.50 77.85  0.66 78.92  0.77

Shrub Canopy Canopy Canopy Ground Cavity Canopy Emergent Canopy Canopy Cavity Cavity Canopy Shrub Canopy Ground Canopy Ground Ground Shrub Ground Canopy Canopy Ground Canopy Ground Cavity Ground Ground Canopy Ground Canopy Ground Ground Cavity Ground Canopy Ground Ground Canopy Canopy Canopy Cavity Canopy Emergent Canopy Emergent Ground Emergent Canopy Ground Ground Canopy Emergent Ground Cavity Canopy Canopy Ground Cavity Ground Canopy Emergent Shrub

a

Riparian species.

significant differences resulted between treatments for canopy closure (Axis 1; F3,72 = 54.26, p < 0.001), non-riparian grass cover (Axis 2; F3,72 = 5.06, p < 0.005) and canopy height (Axis 3; F3,72 = 4.72, p < 0.01). Pair-wise comparisons of treatments indi-

cated significant differences between burned and harvested treatments for canopy closure (all ps < 0.001). Non-riparian grass cover differed between the burned treatments (p < 0.005) and canopy height differed only between burned non-merchantable

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

60

and 30 m buffer treatments (p < 0.005). Significant differences in PCA scores were also found between treatments in percent of postdisturbance meadow marsh (Axis 1; F3,81 = 11.56, p < 0.001) with pair-wise comparisons showing differences between both burned treatments and the 10 m buffer treatments (both ps < 0.001). All significant post-disturbance vegetation variables were used in subsequent avian analyses. 3.3. Avian Twenty-three species differed in relative abundance and frequency of occurrence between treatments using Indicator Species Analysis (Table 4). Two riparian species, the Common Yellowthroat and Eastern Kingbird (Tyrannus tyrannus) were indicative of burned merchantable sites. Eight species of upland birds were indicators of burned merchantable sites including American Robin (Turdus migratorius), Brown-headed Cowbird (Molothrus ater), Clay-colored Sparrow (Spizella pallida), Connecticut Warbler (Oporornis agilis), House Wren (Troglodytes aedon), Least Flycatcher (Empidonax minimus), Philadelphia Vireo (Vireo philadelphicus) and Western Wood-Pewee (Contopus sordidulus). The indicator value for Le Conte’s Sparrow (Ammodramus leconteii) was highest in burned non-merchantable sites; however, it also had a high indicator value in burned merchantable sites suggesting a positive response to pyrogenic riparian habitats. No upland species were indicators of the burned non-merchantable treatment. Alder Flycatcher (Empidonax alnorum) and Wilson’s Warbler (Wilsonia pusilla) were riparian species indicative of 30 m buffers and no riparian species were indicative of 10 m buffers. The Chipping Sparrow (Spizella passerina), Cape May Warbler (Dendroica tigrina), Gray Jay (Perisorius canadensis) and Pine Siskin Table 4 Species with significant (p < 0.1) differences in Indicator Values (combining both abundance and frequency of occurrence) in: (1) burned merchantable (B1; n = 21), (2) burned non-merchantable (B2; n = 29), (3) 10 m buffers (H1; 25% retention in the next 30 m; n = 18), and (4) 30 m buffers (H2; n = 21) in the Boreal Plain of Saskatchewan and Alberta, 2004 and 2005 determined using Indicator Species Analysis (ISA) Maximum

Indicator value

Group

Species

B1

B2

H1

H2

Burned merchantable

American Robin Brown-headed Cowbird Clay-colored Sparrow Connecticut Warbler Common Yellowthroata Eastern Kingbirda House Wren Least Flycatcher Philadelphia Vireo Western Wood-Pewee

17 30 33 11 21 21 22 29 18 22

2 4 4 0 1 1 0 0 0 7

2 0 1 1 0 2 0 4 1 0

6 1 1 0 6 1 0 1 1 1

Burned non-merchantable

Le Conte’s Sparrowa

15

19

0

2

10 m buffer

Chipping Sparrow Cape May Warbler Gray Jay Pine Siskin

12 0 11 0

7 0 0 0

29 15 18 12

12 1 3 7

30 m buffer

Alder Flycatchera Black-and-White Warbler Chestnut-sided Warbler Magnolia Warbler Rose-breasted Grosbeak Swainson’s Thrush Tennessee Warbler Wilson’s Warblera

14 0 2 0 0 1 3 0

2 2 0 0 0 1 1 0

4 1 0 0 0 4 22 1

22 12 18 29 19 19 24 20

Scientific names for species are shown in Table 3. Bold values indicate the treatment in which the species was most abundant/frequently occurring. a Riparian indicator species.

(Carduelis pinus) were upland birds indicative of 10 m buffers and Black-and-White Warbler (Mniotilta varia), Chestnut-sided Warbler (Dendroica pennsylvanica), Magnolia Warbler (Dendroica magnolia), Rose-breasted Grosbeak (Pheucticus ludovicianus), Swainson’ Thrush (Catharus ustulatus) and Tennessee Warbler (Vermivora peregrina) were upland species indicative of 30 m buffers. Treatments showed considerable variability and separation between burned and harvested sites in the overall Redundancy Analysis ordination with some overlap between post-fire and postharvest sites based on 67% Confidence Ellipses (Fig. 3). High overlap was exhibited within burned treatments and within harvested treatments, but there was little overlap between postfire and post-harvest bird communities. Both harvest treatments had similar bird community composition, and 30 m buffers exhibited greater variability in community composition than 10 m buffers. Results from MRPP confirmed that overall bird communities associated with each treatment were significantly different (T = 7.598, A = 0.023, p < 0.0001). Pair-wise comparisons using MRPP were all significant with small but significant differences for the comparison between the burned treatments (T = 2.796, A = 0.009, p < 0.05) and the two harvest treatments (T = 2.403, A = 0.011, p < 0.05). The first four axes of the overall RDA explained 14.2% of the variance in the species matrix. The first two axes explained 27.2% and 20.5%, respectively, of the species–environment relationship. We interpreted the first axis as representing a gradient from closed- to open-canopy habitats. Axis 2 was positively associated with area disturbed and negatively associated with hardwood vegetation; hence, we interpreted this axis as representing a gradient from high to low disturbance. Riparian species positively associated with the vectors representing burned non-merchantable habitat and percent disturbed include Greater Yellowlegs (Tringa melanoleuca), House Wren and Swamp Sparrow (Melospiza georgiana). Fewer species were associated with the burned nonmerchantable vector. In general, more species considered forest specialists including many warbler species were within the Confidence Ellipses of harvested treatments (e.g., Black-throated Green Warbler Dendroica virens, Cape May Warbler). Species typically found in softwood forest types were also associated with the harvested sites including Pine Siskin (Carduelis pinus), Tennessee Warbler and Bay-breasted Warbler (Dendroica castanea). Riparian species associated positively with harvested treatments were Wilson’s Warbler and Orange-crowned Warbler (Dendroica celata). Another riparian species, the Northern Waterthrush (Seiurus noveboracensis), was weakly associated with canopy closure and hardwood forest. RDA for riparian species showed 67% Confidence Ellipse around samples in each treatment with greater overlap of treatments compared to the overall ordination (Fig. 4). The riparian bird community showed lower variation with shorter ordination axes than the overall community. High overlap existed between postfire sites and between post-harvest sites. Ten meter buffers showed a tighter cluster in the ordination being positioned mostly within the Confidence Ellipses of 30 m buffers. Results from MRPP showed significant differences between bird communities associated with each treatment (T = 3.56, A = 0.046, p < 0.005) with pair-wise comparisons indicating only burned non-merchantable sites and 10 m buffers were significantly different (T = 3.401, A = 0.024, p < 0.01). The first two axes of the RDA represented 19.8% of variance in the riparian species data and 59.4% of the species–environment matrix. The first axis encompassed 7.4% of the total variance in the species data and 39.4% of the total variance in the species environment relationship. This axis was positively associated with

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

61

Fig. 3. Redundancy Analysis (RDA) ordination plot with 67% Confidence Ellipses (CE) for all bird (upland and riparian) species (71 species) in four treatments: (1) Burned merchantable (B1; n = 21), (2) Burned non-merchantable (B2; n = 29), (3) 10 m buffer (H1; 25% retention in the next 30 m; n = 18), and, (4) 30 m buffer (n = 21) around small wetlands in the Boreal Plain of Saskatchewan and Alberta, 2004 and 2005. Species codes and scientific names for species are listed in Table 3. Vectors indicate strength (length) and direction of relationship between species and explanatory variables. Pre-disturbance vegetation data used as covariates include area of low merchantable softwood, graminoid and hardwood vegetation. Post-disturbance vegetation variables used as explanatory variables include area disturbed, area of fen and canopy closure. Refer to text for details on vegetation analyses. Riparian species are marked with and asterisk.

graminoid vegetation, which was positively correlated with the burned merchantable treatment. Axis 1 was negatively associated with canopy closure and softwoods. Similar to the overall ordination, we interpreted the first axis as representing a gradient from upland to lowland habitat types. Species showing a strong positive association with Axis 1 were Le Conte’s Sparrow, Yellowheaded Blackbird (Xanthocephalus xanthocephalus) and Eastern Kingbird. Orange-crowned Warbler had the highest negative association with Axis 1. Axis 2 represented 2.0% of the variance in the bird data and 20.0% of the variance in the species– environment relationship. This axis was positively associated with area of hardwood vegetation and negatively associated with area of fen; hence, we interpreted this axis as representing a gradient from merchantable to non-merchantable forest types. Burned nonmerchantable sites were negatively correlated with this axis and highly correlated with fen. Swamp Sparrow showed the strongest positive association with Axis 2 with Olive-sided Flycatcher (Contopus cooperi) and Northern Waterthrush showing weaker associations with this axis. Solitary Sandpiper (Tringa solitaria) and Sora Rail (Porzana carolina) showed weak, negative associations with Axis 2.

4. Discussion While previous studies have investigated the response of upland birds to forestry and fire (see Schieck and Song, 2006), this study is the first to contrast the influence of natural (fire) and anthropogenic (forest harvesting) disturbances in shoreline forests and riparian areas on bird communities at the riparian ecotone. Due to the distribution of our selected treatments (e.g., within 4 fires and 20 cutblocks within two Forest Management Areas with different riparian management guidelines) on the landscape, our samples were not randomly interspersed. In an attempt to reduce the possibility of pseudoreplication, we sampled as many fires and cutblocks as possible that met our criteria. However, there is potential that our data were influenced by differences in geographical area, pre-disturbance vegetation composition, site productivity or bird community composition. Since we were not able to directly assess between-treatment differences in pre-disturbance vegetation parameters, we used forest inventory data to determine if differences did occur. Results from pre-disturbance vegetation analyses using plot scores from PCA showed that no differences occurred between merchantable

62

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

Fig. 4. Redundancy Analysis (RDA) ordination plot and 67% Confidence Ellipses for riparian-associated species (24 species) in four treatments: (1) Burned merchantable (B1; n = 21), (2) Burned non-merchantable (B2; n = 29), (3) 10 m buffer (H1; 25% retention in the next 30 m; n = 18), and, (4) 30 m buffer (n = 21) in the Boreal Plain of Saskatchewan and Alberta, 2004 and 2005. Species codes and scientific names are listed in Table 3. Vectors indicate strength (length) and direction of relationship between species and explanatory variables. Pre-disturbance vegetation data used as covariates include area of low merchantable softwood, graminoid and hardwood vegetation. Postdisturbance vegetation variables used as explanatory variables include area disturbed, area of fen and canopy closure. Refer to text for details on vegetation analyses.

treatments for all vegetation parameters. Differences occurred among treatments only between the non-merchantable treatment and all other treatments in area of non-merchantable softwood vegetation and between the burned non-merchantable treatment and both harvest treatments for graminoid vegetation. This indicated that the vegetation composition of our selected sites prior to disturbance did not differ and allowed us to contrast bird communities in these treatments post-disturbance. Merchantable treatments showed differences in post-disturbance vegetation structure in area disturbed, canopy closure and meadow marsh. Two riparian species had greater frequency of occurrence and abundance in burned merchantable sites than harvested sites, the Common Yellowthroat and Eastern Kingbird, both generally inhabit open shrubby habitats. Upland species indicative of this treatment were open-habitat generalists (e.g., American Robin, Brown-headed Cowbird, Western Wood-Pewee) or species associated with shrubby or early successional habitats (Clay-colored Sparrow, Least Flycatcher and Philadelphia Vireo); however, one cavity-nesting species, the House Wren, and another species associated with lowland burned habitats, the Connecticut Warbler, were also indicators of this treatment. These results are similar to those found in other studies examining bird–habitat relationships and comparisons of post-fire and post-harvest bird communities (reviewed in Schieck and Song, 2006). The burned non-merchan-

table treatment, which showed the greatest differences in vegetation composition between the treatments, had a single indicator species, the Le Conte’s Sparrow. This species also had a high indicator value in the burned merchantable treatment. The propensity of this species to occupy burned riparian habitats may be due to the greater openness of both the post-fire riparian and upland habitats and to the greater availability of graminoid vegetation in non-merchantable sites. The only riparian indicator species of harvested treatments were the Alder Flycatcher and Wilson’s Warbler. These species are most often found in tall, dense shrubby habitats and their greater abundance in harvested sites may be due to the greater availability of this habitat than in early post-fire habitats. However, vegetation analysis did not show differences between treatments in tall shrub habitats. Upland species that nest in conifer-dominated habitats (Chipping Sparrow, Cape May Warbler, Gray Jay and Pine Siskin) were indicators of the 10 m buffer treatment suggesting that, although differences were not found in tree composition between treatments, more coniferous trees may exist in sites with smaller buffers (i.e., closer to the water’s edge). This is further supported by the result that the species indicative of the 30 m treatment are mostly associated with deciduous or mixedwood forest (e.g., Chestnut-sided Warbler, Magnolia Warbler, Rose-breasted Grosbeak, Swainson’s Thrush). Results

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

from these analyses are consistent with other studies that have examined upland bird responses to fire and forestry (e.g., Hobson and Schieck, 1999; Van Wilgenburg and Hobson, 2008). Redundancy Analysis showed that differences in bird community composition were greater for the overall (upland and riparian) community compared to the analysis with only the riparian species. The overall bird communities in merchantable treatments were significantly different between burned and harvested sites whereas the only differences in the riparian community were between the non-merchantable and the 10 m buffer treatment. Based on the ordination, divergence in the overall bird community appeared to be driven by area of open habitat, canopy closure, graminoid vegetation and softwood forest. Riparian species were associated with burned treatments and/or with graminoid vegetation including Tree Swallow (Tachycineta bicolor), Yellowheaded Blackbird, Palm Warbler (Dendroica palmarum), Lincoln’s Sparrow (Melospiza lincolnii), Common Yellowthroat, Le Conte’s Sparrow and Eastern Kingbird. The last three of these species were also indicator species of burned treatments giving greater confidence in the ordinations. Both burned open and graminoid habitats are likely preferred by Tree Swallows for foraging. Early post-fire upland habitats may act as an extension of the riparian ecotone by providing suitable early successional habitat (e.g., shrubby, no canopy), allowing some riparian species to expand or occupy territories protruding into uplands. Common Yellowthroats and Eastern Kingbirds were indeed detected in regenerating upland post-fire stands adjacent to wetlands lending support to this hypothesis. Only three riparian species including Wilson’s Warbler, Northern Waterthrush and Orange-crowned Warbler were found within the Confidence Ellipses of the harvest treatments and were associated with the vectors representing canopy closure, hardwood and softwood forest, respectively. These species typically inhabit treed swamps with tall shrubby vegetation, which may not be available in early post-fire sites. Lack of vegetation succession in shoreline forests back to early seral stages may limit riparian bird species in harvested landscapes (e.g., by retaining buffers) to the non-forested portion of the riparian habitat by not providing the natural range of habitats found postfire (e.g., dense upland shrubs adjacent to the riparian area with no canopy). Most of the divergence in the overall bird community was due to the upland bird community as shown by comparison to the riparian ordination. Upland species associated with burned sites included those that prefer burned forest (e.g., Black-backed Woodpecker, Brown Creeper, Connecticut Warbler), habitat generalists (e.g., American Robin, Brown-headed Cowbird) and species associated with shrubby habitats (e.g., Clay-colored Sparrow, Least Flycatcher). More canopy-nesting species were associated with harvested treatments than with the burned treatments and responded to canopy closure, softwood and hardwood vegetation parameters. Many warbler species including Cape May, Magnolia, Black-throated Green and Black-and-White Warbler along with the Swainson’s Thrush were associated with the vector representing canopy closure. Greater convergence of riparian bird communities was seen in the Redundancy Analyses. This result occurred even though we compared burned and unburned riparian vegetation and suggests less of an impact of fire and shoreline forest harvesting on riparian versus upland species in the first 4 years post-disturbance. This was unexpected because there were major differences in the riparian vegetation structure of burned (e.g., elimination of shrubby habitat) and harvest treatments (i.e., no change in riparian vegetation structure caused by forestry). Some riparian species did show strong associations with the vectors representing graminoid (e.g., Eastern Kingbird, Yellow-headed Blackbird) and hardwood vegetation (Yellow Warbler, Swamp Sparrow). This suggests that

63

post-fire and post-harvest riparian sites may indeed have some characteristics preferable to some riparian bird species. However, the majority of riparian species showed weak responses to treatment. These results have important implications for shoreline forest management as they indicate that disturbances in the upland do not have much impact on the riparian bird community. Surprisingly, bird communities in burned sites showed greater overlap with larger buffers than with smaller buffers. Larger buffers may provide a greater range of habitats suitable for more riparian species (e.g., a more developed shrub layer, more canopy cover) in the first 4 years post-harvest. However, smaller buffers may provide habitats more similar to riparian areas in later stages of early succession (5–10 years) with regenerating cutblocks acting as a surrogate for riparian habitat (Hobson and Schieck, 1999). Harvested treatments did not include any sites that had the entire buffer removed compared to burned sites where most trees adjacent to the riparian area were burned. Wetlands with the adjacent buffer entirely harvested may provide successional habitats for riparian species most similar to burned shoreline forests, and future studies should examine this as a potential treatment. Hypotheses regarding natural disturbance approximation focus on post-fire structural attributes, patterns of disturbance and temporal rates of disturbance. This study addressed only the structural aspect of the Natural disturbance hypothesis of post-fire riparian and shoreline forest bird communities. Post-fire and postharvest vegetation characteristics differed markedly within and between treatments and, therefore, patterns and rates of natural disturbances should also be considered when implementing practices based on results from this study. To fully understand bird community-disturbance dynamics at the riparian interface and to more accurately implement shoreline forest management strategies, more research should be conducted on understanding differences in: (1) bird community composition between burned and unburned shoreline forests, and (2) between burned upland and burned shoreline forests. If there are no apparent differences in burned and unburned riparian bird communities, then shoreline forest management may not be an issue for riparian bird conservation. Such studies should include forests at various seral stages post-fire at different burn severities to include the full possible suite of potential bird communities associated with shoreline forest age and structure. Secondly, if burned shoreline forest was found to be important for upland birds (e.g., more diverse or rich bird community, greater productivity) compared to upland forest, then management of shoreline forests should include leaving these areas to burn, either naturally or through prescribed burning. 5. Conclusions Natural disturbances are critical components of the ecology of the boreal forest and to the maintenance of the species that coexist with them (Potter and Kessell, 1980; Brawn et al., 2001). Current shoreline forest harvesting guidelines were not developed originally to approximate natural disturbances. Management policies typically mandate the retention of one-sized treed buffers to mitigate the potential negative impacts of forestry on water quality or for aesthetic purposes (Hickey and Doran, 2004). This study has shown that fire and forestry in boreal shoreline forest and riparian areas result in different bird communities. Overall, upland bird communities in post-fire sites exhibited a higher natural range of variability and showed more differences in composition than riparian bird communities early post-disturbance. Forest managers attempting to maintain natural diversity and bird community composition on the landscape should

64

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65

consider differential bird community responses to fire and buffer treatments by more closely approximating patterns of fires in shoreline forests. Larger buffers may provide habitat for upland species, but may reduce the amount of habitat for riparian birds that would normally be available after disturbance in the shoreline forest, especially during the period when these areas resemble shrubby riparian habitat (e.g., 5–10 years post-disturbance). The retention of larger (30 m) treed buffer strips may be a management scenario suitable for some riparian species, especially those that prefer tall and dense shrub habitats and canopy cover (e.g., Alder Flycatcher, Northern Waterthrush). However, a static approach of retaining buffer strips on all water bodies on the landscape may not satisfy the needs of all riparian species, especially those that prefer shrubby habitats with little or no canopy cover. Some species, including Common Yellowthroat and Lincoln’s Sparrow were detected in regenerating post-fire and post-harvest (1–5 years) uplands and these species may use such habitats as a surrogate for, or as an extension of, riparian habitat. Therefore, cutting closer to water bodies (e.g., harvesting within buffers) may create habitat for some riparian species, especially in later stages of early shoreline forest succession. As such, a uniform prescription of single buffer strip guidelines across an entire landscape may not provide habitat for all riparian species, especially if productivity of these species is higher at the riparian-upland interface than in regenerating upland forests. Riparian bird community analysis showed less difference than the combined upland and riparian analysis. Therefore, species inhabiting riparian areas of open-water wetlands may be less affected by disturbances in the shoreline forest and riparian habitat than upland bird communities. While lack of a major response to disturbance type by riparian species does not justify eliminating riparian buffers on the landscape, it does allow for the exploration of alternative management scenarios for shoreline forests as they pertain to riparian species. Forest harvesting is expected to truncate forest age distribution (70 years) in the boreal forest with the majority of old-growth forest remaining in buffer strips. Therefore, species that may be most at risk of population declines due to forest harvesting are those that require old-growth forest (e.g., Black-throated Green Warbler). These species may benefit from forestry practices that take into account the retention of old-growth in larger patches (Schmiegelow et al., 1997). Re-allocation of trees in buffer strips to larger aggregate patches within adjacent cutblocks may allow the creation of more suitable remnant habitats for old-growth dependant species. This approach would require management at a landscape level that includes incorporating riparian areas to ensure that habitat preferences of all species on the landscape are accommodated, including riparian species. The trade-off between managing buffers as static elements of the landscape as opposed to dispersing the same volume of timber as residual in other portions of the landscape should be further assessed, potentially via simulation studies (e.g., Rempel and Kaufmann, 2003; Rempel et al., 2007). In order to better approximate the natural disturbance process it may be appropriate to burn buffer strips so that they more closely resemble naturally burned shoreline forest and riparian areas not only initially but also during succession (Lilja et al., 2005). This management approach assumes that shoreline forest and riparian areas left in buffer strips would not burn as frequently or at the same severity as similar areas in unmanaged landscapes. This may indeed be the case with increasing fragmentation of the boreal forest from oil and gas development, agricultural conversion and with the expanding road network associated with these developments. However, if riparian buffers represent the major component of oldgrowth remaining on the landscape, this is not a suitable option.

Using a flexible template for harvesting shoreline forests based on the natural disturbance hypothesis may be a step closer to approximating natural disturbances for riparian birds. This approach could include eliminating buffers or applying various widths of buffers on single wetlands or on different wetlands across the landscape to encompass the natural range of variability of birds in post-fire environments. However, these suggestions are speculative and more research is needed to investigate early postfire riparian and shoreline forest vegetation composition and structure. Such research should quantify residual patch size and composition at the riparian edge with different shoreline forest and wetland types and bird responses to these variables relative to similar configurations in harvested areas. Future studies should also contrast bird community composition between disturbed and undisturbed riparian areas and between post-fire upland and shoreline forests. This information would be useful for coarse-filter forest management in understanding how and if riparian bird communities are actually affected by disturbances in the upland and if post-fire shoreline forests provide more suitable habitats for upland bird communities. Acknowledgements Drs. S. Hannon, B. Clark and K. Van Rees and two anonymous reviewers gave comments that greatly improved this manuscript. This project was funded by the Alberta Conservation Association (project # 0308090101), the Saskatchewan Forest Development Fund (project # 200413, Saskatchewan Forest Center), the Sustainable Forest Management Network (SFMN, project # hobsonkbore9) and the Western Boreal Conservation Initiative (Environment Canada). Personal funding was provided to K.J.K. by Saskatchewan Environment, SFMN and the Saskatchewan Wildlife Federation. Canadian Wildlife Service (Prairie and Northern Wildlife Research Center), Ducks Unlimited Canada, Mistik Management Ltd. and Weyerhaeuser Canada provided in-kind and logistical support. L. Best, H. Blair, J. Karst, T. Lanson, S. Prentice and I. Yukes were assiduous field assistants. References Acton, D.F., Padbury, B.A., Stushnoff, C.T., 1998. The Ecoregions of Saskatchewan. Canadian Plains Research Center, Regina, Saskatchewan, p. 205. Alldredge, M.W., Pollock, K.H., Simons, T.R., Shriner, S.A., 2007. Multiple species analysis of point count data: a more parsimonious modeling framework. Journal of Applied Ecology 44, 281–290. Armstrong, G.W., Adamowicz, W.L., Beck, J.A., Cumming, S.G., Schmiegelow, F.K.A., 2003. Coarse filter ecosystem management in a non-equilibrating forest. Forest Science 49, 209–223. Attiwill, P.M., 1994. The disturbance of forest ecosystems—the ecological basis for conservative management. Forest Ecology and Management 63, 247–300. Blancher, P., 2003. Importance of Canada’s Boreal Forest to Landbirds. Canadian Boreal Initiative and Boreal Songbird Initiative, Ottawa, Ont. and Seattle, WA, p. 42. Brawn, J.D., Robinson, S.K., Thompson III, F.R., 2001. The role of disturbance in the ecology and conservation of birds. Annual Review of Ecology and Systematics 32, 251–276. Buckland, S.T., Anderson, D.R., Burnham, K.P., Laake, J.L., Borchers, D.L., Thomas, L., 2001. Introduction to Distance Sampling. Oxford University Press Inc., New York, NY, p. 452. Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach, second ed. Springer-Verlag, New York, NY, p. 488. Cumming, S.G., Burton, P.J., Prahacs, S., Garland, M.R., 1994. Potential conflicts between timber supply and habitat protection in the boreal mixedwood of Alberta, Canada: a simulation study. Forest Ecology and Management 68, 281– 302. Darveau, M., Beauchesne, P., Be´langer, L., Huot, J., LaRue, P., 1995. Riparian forest strips as habitat for breeding birds in the boreal forest. Journal of Wildlife Management 59, 67–78. Decamps, H., Pinay, G., Naiman, R., Petts, McClain, M.E., Hillbricht-Ilkowska, A., Hanley, T.A., Holmes, R.M., Quinn, J., Gibert, J., Planty Tabacchi, A.-M., Schiemer, F., Tabacchi, E., Zalewski, M., 2004. Riparian zones: where biogeochemistry meets biodiversity in management practice. Polish Journal of Ecology 52, 3–18.

K.J. Kardynal et al. / Forest Ecology and Management 257 (2009) 54–65 DeLong, S.C., 2002. Using nature’s template to best advantage in the Canadian boreal forest. Silva Fennica 36, 401–408. DeLong, S.C., Tanner, D., 1996. Managing the pattern of forest harvest: lessons from wildfire. Biodiversity and Conservation 5, 1191–1205. Dufrene, M., Legendre, P., 1997. Species assemblages and indicator species: the need for a flexible asymmetrical approach. Ecological Monographs 67, 345–366. Dunn, E.H., 2002. Using declines in bird populations to identify needs for conservation action. Conservation Biology 16, 1632–1637. Farnsworth, G.L., Pollock, K.H., Nichols, J.D., Simons, T.R., Hines, J.E., Sauer, J.R., 2002. A removal model for estimating detection probabilities from point-count surveys. Auk 119, 414–425. Gregory, S.V., Swanson, F.J., McKee, W.A., Cummins, K.W., 1991. An ecosystem perspective of riparian zones. BioScience 41, 540–551. Hannon, S.J., Paszkowski, C.A., Boutin, S., DeGroot, J., Macdonald, S.E., Wheatley, M., Eaton, B.R., 2002. Abundance and species composition of amphibians, small mammals, and songbirds in riparian forest buffer strips of varying widths in the boreal mixedwood of Alberta. Canadian Journal of Forest Research 32, 1784– 1800. Harper, K.A., Macdonald, S.E., 2001. Structure and composition of riparian boreal forest: new methods for analyzing edge influence. Ecology 82, 649–659. Harris, A.G., McMurray, S.C., Uhlig, P.W.C., Jeglum, J.K., Foster, R.F., Racey, G.D., 1996. Field Guide to the Wetland Ecosystem Classification for Northwestern Ontario. Ontario Ministry of Natural Resources, Northwest Science & Technology, Thunder Bay, Ontario, Field Guide FG-01, p. 74. Hickey, M.B.C., Doran, B., 2004. A review of the efficiency of buffer strips for the maintenance and enhancement of riparian ecosystems. Water Quality Research Journal of Canada 39, 311–317. Hobson, K.A., Bayne, E., 2000. Breeding bird communities in boreal forest of western Canada: consequences of ‘‘unmixing’’ the mixedwoods. Condor 102, 759–769. Hobson, K.A., Schieck, J., 1999. Changes in bird communities in boreal mixedwood forest: harvest and wildfire effects over 30 years. Ecological Applications 9, 849–863. Huggins, R.M., 1989. On the statistical analysis of capture–recapture experiments. Biometrika 76, 133–140. Hunter, M.L., 1993. Natural fire regimes as spatial models for managing boreal forests. Biological Conservation 65, 115–120. Imbeau, L., Savard, J.P.L., Gagnon, R., 1999. Comparing bird assemblages in successional black spruce stands originating from fire and logging. Canadian Journal of Zoology 77, 1850–1860. Kirk, D.A., Diamond, A.W., Smith, A.R., Holland, G.E., Chytyk, P., 1997. Population changes in boreal forest birds in Saskatchewan and Manitoba. Wilson Bulletin 109, 1–27. Laake, J., 2007. RMark version 1.6.4. Website: http://www.r-project.org/. Lee, P.C., Crites, S., Nietfeld, M., Van Nguyen, H., Stelfox, J.B., 1997. Characteristics and origins of deadwood material in aspen-dominated boreal forests. Ecological Applications 7, 691–701. Legendre, P., 1993. Spatial autocorrelation: trouble or new paradigm? Ecology 74, 1659–1673. Lilja, S., De Chantal, M., Kuuluvainen, T., Vanha-Majamaa, I., Puttonen, P., 2005. Restoring natural characteristics in managed Norway spruce [Picea abies (L.) Karst.] stands with partial cutting, dead wood creation and fire: immediate treatment effects. Scandinavian Journal of Forest Research 20 (Suppl. 6), 68–78. Macdonald, E., Burgess, C.J., Scrimgeour, G.J., Boutin, S., Reedyk, S., Kotak, B., 2004. Should riparian buffers be part of forest management based on emulation of natural disturbance? Forest Ecology and Management 187, 185–196. Martell, K.A., Foote, A.L., Cumming, S.G., 2006. Riparian disturbance due to beavers (Castor canadensis) in Alberta’s boreal mixedwood forest: implications for forest management. Ecoscience 13, 164–171. McCune, B., Grace, J.B., 2002. Analysis of Ecological Communities. MjM Software Design, Gleneden Beach, Oregon, USA.

65

McCune, B., Mefford, M.J., 1999. PC-ORD. Multivariate analysis of Ecological Data version 4.20. MjM Software, Gleneden Beach, Oregon, USA. Naiman, R.J., Decamps, H., Pollock, M., 1993. The role of riparian corridors in maintaining regional biodiversity. Ecological Applications 3, 209–212. National Wetlands Working Group, 1997. In: Warner, B.G., Rubec, C.D.A. (Eds.), The Canadian Wetland Classification System, second edition. Wetlands Research Centre, University of Waterloo, Waterloo, Ontario. Potter, M.W., Kessell, S.R., 1980. Predicting mosaics and wildlife diversity resulting from fire disturbance to a forest ecosystem. Environmental Management 4, 247–254. Quinn, G.P., Keough, M.J., 2004. Experimental Design and Data Analysis For Biologists. Cambridge University Press, Cambridge, UK, p. 560. R Development Core Team, 2007. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Rempel, R.S., Baker, J., Elkie, P.C., Gluck, M.J., Jackson, J., Kushneriuk, R.S., Moore, T., Perera, A.H., 2007. Forest policy scenario analysis: sensitivity of songbird community to changes in forest cover amount and configuration. ACE-E´CO 2, 5. [online] URL: http://www.ace-eco.org/vol2/iss1/art5/. Rempel, R.S., Kaufmann, C.K., 2003. Spatial modeling of harvest constraints on wood supply versus wildlife habitat objectives. Environmental Management 32, 334– 347. Rowe, J.S., Scotter, G.W., 1973. Fire in the boreal forest. Quaternary Research 3, 444– 464. Sabo, J.L., Sponseller, R., Dixon, M., Gade, M., Harms, T., Heffernan, J., Jani, A., Katz, G., Soykan, C., Watts, J., Welter, J., 2005. Riparian zones increase regional richness by supporting different, not more species. Ecology 86, 56–62. Saskatchewan Environment, 2006a. Forest Management Area Agreement Standards and Guidelines. Mistik Management Ltd., Forest Ecosystem’s Branch, Regina, SK. Saskatchewan Environment, 2006b. Weyerhaeuser Prince Albert Forest Management Area Agreement Standards and Guidelines. Forest Ecosystem’s Branch, Regina, SK. Schieck, J., Song, S.J., 2006. Changes in bird communities throughout succession following fire and harvest in boreal forests of western North America: literature review and meta-analyses. Canadian Journal of Forest Research 36, 1299– 1318. Schmiegelow, F.K.A., Machtans, C.S., Hannon, S.J., 1997. Are boreal birds resilient to forest fragmentation? An experimental study of short-term community responses. Ecology 78, 1914–1932. Simon, N.P.P., Schwab, F.E., Otto, R.D., 2002. Songbird abundance in clear-cut and burned stands: a comparison of natural disturbance and forest management. Canadian Journal of Forest Research 32, 1343–1350. Smith, A.R., 1993. Ecological profiles of birds in the boreal forest of western Canada. In: Juhnke, H. (Ed.), Birds in the Boreal Forest. Forestry Canada, Northern Forestry Centre, Edmonton, Alberta, Canada, pp. 37–38. SPSS for Windows, Release 14.0.0., 2005. SPSS Inc., Chicago. ter Braak, C.J.F., 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67, 1167– 1179. ter Braak, C.J.F., Smilauer, P., 1999. CANOCO Reference Manual and User’s Guide to Canoco for Windows. In: Software for Community Ordination Version 4.54, Microcomputer Power, Ithaca, New York. Timoney, K.P., 2003. The changing disturbance regime of the boreal forest of the Canadian Prairie Provinces. The Forestry Chronicle 79, 502–516. Van Wilgenburg, S.L., Hobson, K.A., 2008. Landscape-scale disturbance and boreal forest birds: can large single-pass harvest approximate fires? Forest Ecology and Management 256, 136–146. Welsh, D.A., 1987. The influence of forest harvesting on mixed coniferous-deciduous boreal bird communities in Ontario, Canada. Acta Oecologia 8, 247–252. White, G.C., 2006. Program MARK. Mark and Recapture Survival Estimation. Department of Fishery and Wildlife, Colorado State University.