1997 Schmiegelow Birds Resilient Fragmentation Short-Term

Ecology, 78(6), 1997, pp.
1914–1932
q 1997 by the Ecological Society of America
ARE BOREAL BIRDS RESILIENT TO FOREST FRAGMENTATION? AN

EXPERIMENTAL STUDY OF SHORT-TERM COMMUNITY RESPONSES
FIONA K. A. SCHMIEGELOW,1 CRAIG S. MACHTANS,2 AND SUSAN J. HANNON2
1Department of Zoology, University of British Columbia, 6270 University Boulevard, Vancouver,
British Columbia, Canada V6T 1Z4
2Department of Biological Sciences, University of Alberta, Edmonton, Alberta, Canada T6G 2E9
Abstract. We studied the effect of habitat fragmentation on the richness, diversity,

turnover, and abundance of breeding bird communities in old, boreal mixed-wood forest
by creating isolated and connected forest fragments of 1, 10, 40, and 100 ha. Connected
fragments were linked by 100 m wide riparian buffer strips. Each size class within treatments
and controls was replicated three times. We sampled the passerine community using point
counts before, and in each of two years after, forest harvesting, accumulating 21 340 records
representing 59 species. We detected no significant change in species richness as a result
of the harvesting, except in the 1-ha connected fragments, where the number of species
increased two years after isolation. This increase was accounted for by transient species,
suggesting that the adjacent buffer strips were being used as movement corridors. Diversity
(log series alpha index) was dependent on area in the isolated fragments only after cutting,
having decreased in the smaller areas. Turnover rates in the isolated fragments were sig-
nificantly higher than in similar connected or control areas, due to species replacement.
Crowding occurred in the isolated fragments immediately after cutting, but two years after
fragmentation, the responses in abundance of species varied with migratory strategy. Num-
bers of Neotropical migrants declined in both connected and isolated fragments, and resident
species declined in isolated fragments. Most species in these groups require older forest,
many favoring interior areas. Abundance of short-distance migrants, most of which are
habitat generalists, did not change. Overall, although there was no decrease in species
richness from our recently fragmented areas, community structure was altered; maintaining
connections between fragments helped to mitigate these effects. Nevertheless, the magnitude
of the fragmentation effects we documented is small compared with those observed else-
where. Birds breeding in the boreal forest, where frequent small- and large-scale natural
disturbances have occurred historically, may be more resilient to human-induced habitat
changes, such as those caused by forest harvesting. However, these results should be in-
terpreted with caution. First, they are short-term and address only broad-scale community
responses based on species richness and relative abundance. Second, the study area was
embedded in a landscape where large areas of old, mixed forest are still available, potentially
dampening any local-scale impacts of fragmentation.
Key words: boreal mixed-wood forest; connectivity; experimental fragmentation; Neotropical mi-
grants; regional forests; songbird communities.
INTRODUCTION patches of older forest are surrounded by regenerating

forest (McGarigal and McComb 1995, Hagan et al.
Habitat loss and concomitant fragmentation are con-
1996).
cerns central to the conservation of biodiversity. De-
spite efforts to slow rates of habitat destruction, most Both the theory of island biogeography (MacArthur
ecosystems are becoming increasingly fragmented. In and Wilson 1963, 1967) and metapopulation theory
North America, much attention has focused on frag- (Levins 1970) predict species loss from habitat frag-
mentation of forested habitat, particularly where there ments, because of higher extinction and lower reco-
has been permanent forest loss due to agricultural and lonization probabilities in isolated habitats. Island bio-
urban expansion. Although recurrent negative impacts, geography theory also predicts that extinction proba-
such as some edge effects, increased predation and her- bility will vary inversely, and recolonization directly,
bivory, and failure of metapopulation dynamics, have with island area, and that species–area curves from
been identified (Simberloff 1994), there have been few fragmented habitats will have steeper slopes and lower
studies of fragmentation in managed landscapes where intercepts than curves from continuous habitat. Meta-
population theory places greater emphasis on the role
Manuscript received 15 May 1996; revised 29 October
of the intervening matrix in mediating these rates. Con-
1996; accepted 1 November 1996; final version received 2 necting areas with corridors to facilitate movement
December 1996. among patches has been suggested many times as a
1914
September 1997 BOREAL BIRDS AND FOREST FRAGMENTATION 1915
conservation solution (e.g., Mann and Plummer 1993), STUDY AREA

but few empirical studies have supported this recom- We conducted this research near Calling Lake, in
mendation (Hobbs 1992, Simberloff et al. 1992). Al- north-central Alberta, Canada (558 N, 1138 W). Our
though recent evidence suggests that birds use corridors study area encompassed ø140 km2 of boreal mixed-
for adult and juvenile dispersal (Haas 1995, Machtans wood forest, within the moist mixed-wood subregion.
et al. 1996), whether this reduces species loss in ad- Trembling aspen (Populus tremuloides), balsam poplar
joining habitat fragments is not known. (Populus balsamifera), and white spruce (Picea glau-
The boreal forest is the most extensive forest eco- ca) are the most abundant upland tree species in this
system in North America, and probably the least un- region, often occurring together in old, mixed stands,
derstood. The mixed-wood region of the boreal forest whereas black spruce (Picea mariana) characterizes
in Canada extends from northeastern British Columbia wetter sites (Strong and Leggat 1981). The dominant
into southern portions of the Northwest Territories, and understory shrubs are alder species (Alnus tenulfolia,
through Alberta and Saskatchewan into southwestern A. crispa), with lesser amounts of willow (Salix spp.).
Manitoba (Rowe 1972). Until recently, the boreal Various fruiting shrubs (Rubus, Ribes, Rosa spp.), sar-
mixed-wood remained in a relatively natural state, saparilla (Aralia nudicaulis), and other herbaceous
where large, natural disturbances, such as fire and in- plants dominate the lower strata.
sect outbreaks, and smaller scale disturbances, such as Mean summer (early June through mid-August) pre-
treefalls, created a mosaic of stand types and succes- cipitation in the region is ø320 mm, accounting for
sional stages. This naturally patchy habitat has one of .70% of the total yearly precipitation; July is generally
the highest levels of bird species diversity in North the wettest month. The mean summer temperature is
America (Robbins et al. 1986). However, increasing 12.08C, and the mean freeze-free period is 85 d (Strong
pressure from industrial forestry is resulting in wide- and Leggat 1981).
spread habitat fragmentation and changes in forest
METHODS
composition (Schmiegelow and Hannon 1993). Older
forests are being harvested first, and the structural and Experimental design
compositional complexity of these areas has been cor- Our design involved two treatments: isolated and
related with high species diversity and specialization connected forest fragments, with common controls.
(Stelfox 1995), leading to predictable conflicts between Isolates were created by clear-cut logging a 200 m wide
timber production and habitat conservation (Cumming strip around forest patches. Connected patches of forest
et al. 1994). were isolated by 200 m of clear-cutting on three sides,
The goal of our analyses was to estimate the severity with the fourth side connected to 100 m wide riparian
of fragmentation effects on breeding boreal bird com- buffer strips (Fig. 1). Isolated forest fragments were 1,
munities in old, mixed stands. We present the results 10, 40, and 100 ha in size; connected fragments were
from a replicated experiment in which boreal forest was 1, 10, and 40 ha in size. We did not include any 100-ha
harvested to leave older forest fragments of different connected fragments because sufficient, suitable forest
sizes: some completely isolated and some connected adjacent to riparian areas was not available. Controls
by riparian buffer strips. We used birds to monitor the were placed within ø4000 ha of continuous, adjacent
effects of fragmentation because they are relatively forest. Each size class was replicated three times, with-
easy to census, have known sensitivities to habitat frag- in each treatment and control, as suggested by a priori
mentation elsewhere, and are good biological indica- power analyses (Schmiegelow and Hannon 1993).
tors for this system (see Schmiegelow and Hannon We used forest inventory information and extensive
1993). We tested the following predictions: ground-truthing to identify candidate study sites and
1) Species loss from fragmented areas will result in to establish permanent sampling stations in 1992. All
sites were in old (80–130 yr-old), aspen-dominated for-
species–area curves with steeper slopes and lower in-
est, similar in canopy height, canopy closure, tree spe-
tercepts relative to similar-sized areas within contin-
cies composition, and understory features. Variation in
uous forest.
these features was stratified across replicate groups and
2) Small fragments, in particular, will experience
size classes, within treatments and controls. For ex-
higher species turnover than large fragments, and will ample, replicates in group 1 occurred in the youngest
lose old-forest specialists and area-sensitive species. forest (80–100 yr old); replicates in group 2 were in
3) Abundances of certain species will temporarily older, relatively pure aspen forest (90–130 yr old), and
increase in recently fragmented areas, because of dis- replicates in group 3 were in older aspen forest (90–
placement of individuals from adjacent harvested ar- 130 yr old) with some white spruce in the canopy. Each
eas. replicate group was represented by one site in each size
4) Bird communities in connected fragments will class, in each treatment or control. This design allowed
be less affected by adjacent harvesting than will those us to separate the effects of fragment area and of var-
in completely isolated fragments. ious attributes of habitat on the bird community.
1916 FIONA K. A. SCHMIEGELOW ET AL. Ecology, Vol. 78, No. 6
FIG. 1. Experimental layout of the Calling Lake Fragmentation Study (after Schmiegelow and Hannon 1993).
Prefragmentation data were collected in 1993 and 10-d intervals, from the third week in May through late
the study area was harvested in the winter of 1993– June each year. Upon arrival at a station, observers
1994, between November and March, according to the would wait for 1 min and would then record all indi-
experimental design (Fig. 2). Two years of postfrag- vidual birds seen and heard during a 5-min sampling
mentation data (1994 and 1995) were collected. interval, within 50-m and 100-m distance classes. All
records at each station were mapped and movements
Bird sampling were noted, ensuring that individuals were recorded
Permanent sampling stations were spaced at 200-m only once per visit (see Ralph et al. 1993). Care was
intervals, along transects 200 m apart, with the excep- also taken, both in the field and when compiling data,
tion of 2–40 ha connected fragments, where sampling to avoid recording the same individual at adjacent sta-
stations were spaced at 180-m intervals due to anom- tions. Any additional behavioral observations, such as
alies in the shape of these sites. Sampling intensity was an adult carrying nesting material (see Table 1), were
proportional to area: 1-ha sites had one station, 10-ha also recorded on the data cards.
sites had two, 40-ha sites had eight, and 100-ha sites All surveys were conducted between sunrise and
had 20 stations, resulting in a total of 219 stations. In 1000 in the morning, following general standards es-
the 1-ha sites, a 50-m sampling radius was used because tablished by Ralph et al. (1993). Surveys were not con-
of area constraints; a 100-m sampling radius was used ducted if it was raining, nor if estimated wind speed
in all other sites. All comparisons we made were either exceeded 25 km/h (Beaufort level 5; small branches
before/after on the same sites, or between treatments move). The order in which we sampled size classes
and controls in the same size classes; thus, variation within replicate groups and treatments was standard-
in sampling radius between size classes did not bias ized within each survey by sampling treatment and
the analyses. Point count surveys were conducted at control areas in the same replicate group, in the same
each station five times during the breeding season, at order, on the same day. Observer and diurnal variation
FIG. 2. Aerial view of a portion of the study area in November 1993, shortly after harvest. In the foreground is one set
of connected fragments; in the background, one set of isolated fragments. Vertical straight lines are seismic exploration lines;
irregular dark lines in harvested areas are haul roads with cut timber alongside.
were standardized across all stations, within each year, tered on the station and three centered 40 m away at
by rotating observers among study areas and varying 08, 1208, and 2408. Within each plot, ground cover in
the order in which stations in a given area were sam- six categories (all green, forb, shrub, grass, leaf litter,
pled. Eight observers conducted point counts in each moss) was estimated to the nearest 5% in four 1-m2
year, and each round of counts took 2–3 days. Three quadrats; litter depth was measured at four locations;
observers were present in all three years, and an ad- number of stems of shrubs (by species) was recorded
ditional three participated in surveys in two of the three for five 1-m2 quadrats. Numbers of saplings (diameter
years. at breast height, dbh ,2.5 cm) and poles (2.5–8 cm
dbh) of each species were counted in a 0.008-ha nested
Vegetation sampling subplot. The number of trees of each species in four
Detailed vegetation data were collected mid-July diameter classes (8–15, 15–23, 23–38, and .38 cm
through early August at all stations, using a protocol dbh), was recorded for the entire plot. We also recorded
modified from the BBIRD Program (Martin 1992a). At the species, dbh, and height of every snag .12 cm dbh,
each station, four 0.04-ha plots were sampled; one cen- and the total number of snags ,12 cm dbh. We used
clinometers to measure average canopy, subcanopy,
TABLE 1. Abundance weightings for bird observations. and tall shrub height, and measured canopy closure as
the mean of four densiometer readings per plot.
No.
Behavior Weight† detections‡ Data analyses
Singing or countersinging male 1.0 19 885 Prior to analysis, all data sets were tested for nor-
Calling or observed male or female 0.5 1598
Territorial dispute 1.0 8 mality using a Shapiro-Wilks test, and for homogeneity
Pair observed 2.0 146 of variance with the Levene statistic (Conover 1980,
Active nest observed 2.0 23 Norusis 1993). Where possible, data were transformed
Juvenile observed 2.0 15 to satisfy the assumptions of parametric statistical tests.
Adult carrying nesting material 2.0 14
Adult carrying food 2.0 17 When assumptions could not be satisfied, nonpara-
Distraction display 2.0 2 metric tests were used. When analyzing vegetation
† Mean weighting factor 5 0.98. data, we used an alpha level of 0.05, because of large
‡ Total no. detections 5 21 340. sample sizes and associated high power. When inter-
TABLE 2. Summary of vegetation variables and results of analyses of data collected at Calling Lake, Alberta, Canada, 1993–
1995.
Isolated Connected
Variable† Controls fragments fragments F ratio P
Percent cover
All green 52.21 53.45 50.49 1.023 0.361
Grass§ 7.55 7.17 2.72 13.171 ,0.001
Shrubs 25.21 27.10 22.36 2.740 0.067
Leaf litter 71.37 73.38 79.82 (4.019) 0.134
Moss§ 3.92 5.43 5.59 1.526 0.220
Counts
Deciduous stems (no.) 118.14 111.58 102.79 (2.892) 0.058
Deciduous basal area (m2) 3.80 3.64 3.84 0.566 0.569
Mesic coniferous stems (no.) 14.66 9.83 11.48 0.927 0.397
Hygric coniferous stems (no.)§ 2.82 5.54 6.33 0.444 0.642
Coniferous basal area§ (m2) 0.51 0.36 0.58 2.676 0.071
Deciduous saplings (no.) 142.91 101.10 76.52 (15.971) ,0.001
Coniferous saplings (no.)§ 3.03 2.23 1.24 4.037 0.019
Shrub stems (no.) 95.56 93.46 88.67 0.453 0.636
Large snags (no.) 16.34 14.75 13.36 1.491 0.228
Snag basal area (m2) 0.21 0.21 0.15 2.862 0.059
Average snag height (m) 9.87 9.76 9.34 0.163 0.849
Physical descriptors
Canopy height (m) 27.63 27.23 27.08 0.301 0.741
Subcanopy height (m) 13.84 13.50 14.12 0.336 0.715
Tall shrub layer height (m) 3.57 4.08 3.57 4.716 0.010
Canopy closure (m2) 77.83 73.75 74.76 (9.385) 0.009
† All variable values are averages of data at the point-count station level.
‡ All F ratios were calculated using a one-way ANOVA, except for values in parentheses, which are x2 statistics from a
Kruskal-Wallis nonparametric ANOVA. For all tests, n 5 93 for controls and isolated fragments and n 5 33 for connected
fragments; df 5 2, 216.
§ Data were log(x 1 1)-transformed prior to statistical analyses. Nontransformed data are reported in the table.
preting the results of statistical tests on bird community We used one-way ANOVAs to determine if the mean
measures, we used an alpha level of 0.10, because we values of each variable differed between our treatment
feel that Type II errors can be more costly than Type and control areas, because the data did not satisfy the
I errors in applied research (see Peterman 1990, assumptions necessary for MANOVA (Maxwell and
Schmiegelow 1992, Smith 1995). We used the Bon- Delaney 1990). We log(x 1 1)-transformed variables
ferroni correction for multiple comparisons where ap- that were markedly non-normally distributed, and used
propriate (Winer et al. 1991). a Kruskal-Wallis nonparametric ANOVA for variables
Relative sampling area, rather than study site area, that were not homoscedastic between sites.
was used in all regression and ANOVA models. For Description of the bird community.—Although we
1-ha sites, sampled area (one station with a 50-m sam- recorded all bird species seen or heard during point
pling radius) was considered equivalent to actual area, count surveys, we removed all raptors, corvids, shore-
and other sites were scaled accordingly (e.g., a 10-ha birds, waterfowl, and grouse prior to analyses, since
site with two stations, each with a sampling radius of our methods did not sample such species adequately.
100 m, has a sampled area eight times that of a 1-ha We also removed all woodpeckers except the Yellow-
site). bellied Sapsucker (scientific names in Appendix). The
Vegetation patterns.—The original vegetation data Yellow-bellied Sapsucker is the only nonresident
set contained .100 variables. Initially, we used two- woodpecker that occurs in significant numbers in our
dimensional principal component analysis (PCA) to re- study sites. Because this species breeds later, and is
duce the data set. However, the PCA factors did not much more conspicuous than resident woodpeckers in
adequately explain the variation in the data (broken the area, it is better suited to the sampling methods we
stick criteria; Legendre and Legendre 1983), so a subset used. To standardize sampling area and confirm site
of actual variables was chosen for analysis. We used occupancy, we did not include any individuals that
20 variables that either had high factor loadings, or that were only recorded flying over or through the forest
summarized the obvious gradients on the PCA plots. during point counts.
For example (see Table 2), a site–moisture gradient was Bird species richness.—We generated two species
represented by hygric conifers (larch, Larix laricina; richness estimates for each of our study sites in each
black spruce) and mesic conifers (jack pine, Pinus year. One excluded the first round of point counts, con-
banksiana; white spruce). ducted during the third week of May, owing to the high
frequency of late migrants during this period. It oth- in a wide range of circumstances (Southwood 1978),
erwise included all species records, and is herein re- and low sensitivity to variation in sample size (Taylor
ferred to as ‘‘all species.’’ Unless noted, all analyses et al. 1976, Taylor 1978). Taylor (1978) promotes a as
were done using ‘‘all species.’’ The second estimate a good measure of diversity even when the underlying
was based on information from all rounds, but included species abundances do not exhibit a log series distri-
only species for which there was evidence of probable bution. Further, it is less affected by species dominance
or confirmed breeding, following standards similar to than are the more widely used Shannon or Simpson
those used by breeding-bird atlas projects (e.g., Se- indices (Magurran 1988). An additional feature of this
menchuck 1992). We used the following criteria, as- index is that it provides an alternative to rarefaction
sessed at each station: singing male heard on more than for detecting sampling artifacts due to unequal sample
one occasion (i.e., $10 d apart); pair, juvenile, or fam- sizes, as may be the case with some species–area re-
ily group observed; individual observed carrying nest- lationships (e.g., Freemark and Merriam 1986), but it
ing material or food, or performing a distraction dis- preserves more information (Rosenzweig 1995).
play. We refer to this group of species as ‘‘probable We used the Jaccard coefficient (J) to quantify sim-
breeders only’’ (see also Opdam et al. 1985, Hinsley ilarity of sites between years. We used this index be-
et al. 1995). cause it excludes species absent in both samples (see
Species richness and area measures were log-trans- also Krebs 1989) and because it is less sensitive to
formed to generate species–area curves for each treat- variation in sample size and species richness than are
ment and for the controls, in each year. Linear regres- other indices (Wolda 1981). We defined percentage
sion was used to quantify species–area relationships. turnover as (1 2 J) 3 100.
We tested for differences between years within treat- Bird abundance.—All observations were weighted
ment and control groups using repeated-measures anal- according to evidence of breeding (Table 1). Although
ysis of covariance (RM ANCOVA), and for within- it is common to assign a weighting of 2 to singing
year differences between groups (independent obser- males, assuming that a singing male represents a pair
vations) using ANCOVA. (e.g., Helle 1984, Lynch and Whigham 1984), we chose
Measures of diversity and similarity.—We used not to do this for two reasons. First, most detections
abundance curves to characterize overall community during point count surveys are of singing males, re-
structure (May 1975, Southwood 1978) and to select sulting in a mean weighting factor that exceeds 1, and
an appropriate diversity index (Magurran 1988). We thereby artificially inflating the sample. Second, a pos-
pooled data for all species across all treatments and sible outcome of habitat fragmentation is an increase
years for this purpose (Appendix). We first generated in the number of unpaired males (Gibbs and Faaborg
a rank abundance plot (Whittaker 1965), and then gen- 1990, Villard et al. 1993). We also had two reasons for
erated a frequency distribution of species abundances, using mean abundance per sampling round, rather than
plotting number of species against number of individ- maximum abundance in any round (e.g., Welsh and
uals per species (log2 classes, or octaves), after Preston Lougheed 1996), as an estimate of the relative abun-
(1948). Inspection of these curves suggested that the dance of each species at each station in each year. Other
data were best represented by the logarithmic series authors have used the maximum number of birds, be-
(Fisher et al. 1943). We also generated rank abundance cause point count data represent incomplete counts
curves for treatment and control groups, in each year, (Barker and Sauer 1995); in our case, however, this
to check for fits to different models, because we wished could bias results by exacerbating variation due to mul-
to compare communities with a common diversity in- tiple observers, and by equally weighting nonterritorial
dex. All species abundance distributions suggested that or transient individuals.
the most appropriate diversity index for our data was We tested for treatment effects on individual species
the log series alpha (a) (Williams 1964, after Fisher et with more than five detections in any year in both treat-
al. 1943). ments and controls by using simple contrasts, equiv-
Calculation of a values requires two variables: the alent to the interaction term of a two-group repeated-
total number of species in the sample (S) and the total measures ANOVA (von Ende 1993), to compare
number of individuals (N). Through iterative solution changes in abundance over the three study (years 1993–
of Eq. 1, we obtained values for x, the parameter of 1995). Using G tests, we examined the proportion of
the logarithmic series, and then solved Eq. 2 to obtain species that declined after harvesting as a function of
a values for each station and site, in each year: migration strategy.
S (1 2 x) RESULTS
5 (1)
N x(2ln[1 2 x])
Vegetation patterns
N(1 2 x)
a5 . (2) Prior to testing for experimental effects of area and
x
isolation on the bird communities, we first determined
The log series index shows good discriminant ability whether or not our treatments and controls had com-
TABLE 3. Regression statistics for species–area relationships of forest birds in treatment and controls in 1993, 1994, and
1995 (n 5 12 for controls and isolated fragments and n 5 9 for connected fragments, in each year).
Year r2 P z† SE ( z)‡ c§ SE ( c) \
All species
Controls 1993 0.90 ,0.001 0.42 0.042 0.76 0.052
1994 0.81 ,0.001 0.39 0.056 0.76 0.070
1995 0.88 ,0.001 0.38 0.043 0.84 0.054
Isolated fragments 1993 0.78 ,0.001 0.46 0.072 0.70 0.090
1994 0.79 ,0.001 0.42 0.065 0.75 0.081
1995 0.74 ,0.001 0.44 0.077 0.74 0.096
Connected fragments 1993 0.72 0.002 0.40 0.086 0.87 0.080
1994 0.81 ,0.001 0.44 0.075 0.78 0.070
1995 0.74 0.002 0.25 0.051 1.02 0.048
Probable breeders
Controls 1993 0.84 ,0.001 0.42 0.056 0.53 0.070
1994 0.81 ,0.001 0.52 0.075 0.37 0.093
1995 0.86 ,0.001 0.44 0.054 0.51 0.067
Isolated fragments 1993 0.86 ,0.001 0.43 0.052 0.55 0.066
1994 0.75 ,0.001 0.40 0.068 0.60 0.085
1995 0.69 ,0.001 0.44 0.087 0.49 0.108
Connected fragments 1993 0.67 0.004 0.47 0.114 0.55 0.107
1994 0.78 0.001 0.42 0.078 0.60 0.073
1995 0.75 0.002 0.37 0.075 0.62 0.070
† Slope estimate.
‡ Standard error of slope estimate.
§ Intercept estimate.
\ Standard error of intercept estimate.
parable vegetative attributes. Most of the variables used accounting for 63.7% of all records. Short-distance mi-
to describe vegetation structure were similar between grants are second in overall dominance, with 22 species
treatment and control areas (P $ 0.003; df 5 2, 216 representing 30.1% of records; the five resident species
with Bonferroni correction; Table 2). Mean percent included in these analyses account for the remaining
cover of grass was 5% lower in the connected frag- 6.2% of records. We also recorded five additional wood-
ments, a difference we do not consider biologically pecker species (Downy Woodpecker, Picoides pubes-
significant. The density of deciduous saplings (with an cens; Hairy Woodpecker, Picoides villosus; Three-toed
average of 33.0% Alnus spp.) also differed between our Woodpecker, Picoides tridactylus, Black-backed Wood-
treatments and controls; these form a tall shrub layer pecker, Picoides arcticus; and Pileated Woodpecker,
that is an important structural feature of habitats in our Dryocopus pileatus), and five species of corvids (Gray
study area. Alnus density is highly positively correlated Jay, Perisoreus canadensis; Blue Jay, Cyanocitta cris-
with both average yearly abundance and species rich- tata; Black-billed Magpie, Pica pica; American Crow,
ness of birds in all three sites (P # 0.006). However, Corvus brachyrhynchos; and Common Raven, Corvus
when we included density of deciduous saplings as a corax), all residents, over the 3-yr course of this study.
covariate for all tests in the prefragmentation year, we Species–area relationships.—All species–area re-
found that its effect was neglible, so we removed it gressions were significant, with area explaining a large
from subsequent analyses. proportion of the variation in species number across
both treatment and control areas (Table 3), as would
Bird community be expected from our study design. There were no sig-
We analyzed data from 58 passerine species and the nificant changes in either the slope or intercept of these
Yellow-bellied Sapsucker (Appendix). Neotropical mi- relationships within the controls, isolated fragments, or
grant species form the largest component of the com- connected fragments during the three years of this study
munity in old, aspen-dominated stands, with 32 species (1993–1995; Table 4). There were also no significant
TABLE 4. Repeated-measures ANCOVA results for species–area relationships of forest birds within treatment and control
areas 1993–1995 (n 5 12 for controls and isolated fragments and n 5 9 for connected fragments, in each year).
All species Probable breeders only

F ratio df Alpha Beta F ratio df Alpha Beta
Controls 2.55 2, 22 0.101 0.405 2.35 2, 22 0.118 0.437
Isolated fragments 0.06 2, 22 0.942 0.885 1.33 2, 22 0.285 0.623
Connected fragments 1.99 2, 16 0.169 0.511 0.05 2, 16 0.950 0.888
TABLE 5. ANCOVA results for comparison of species–area relationships for forest birds between treatment and control areas
(n 5 12 for controls and isolated fragments and n 5 9 for connected fragments).
Slope† Intercept‡
F ratio Alpha Beta F ratio Alpha Beta
All species
Prefragmentation (1993) 0.24 0.787 0.845 1.51 0.239 0.581
1 yr postfragmentation 0.12 0.888 0.872 0.45 0.643 0.801
Probable breeders only
Prefragmentation (1993) 0.11 0.896 0.874 0.29 0.752 0.834
† df 5 2, 30.
‡ df 5 2, 29.
changes in the relationships among controls, isolated 5 1, 19; P 5 0.943). This effect was most pronounced
fragments, and connected fragments over the same time in small isolates, consistent with our second prediction,
period (Table 5, Fig. 3). These results are contrary to and for resident species. Turnover averaged 92.6% in
our first prediction. the 1-ha isolated fragments, compared with 71.0% and
The only significant change in species richness oc- 66.3% in the 1-ha connected fragments and controls,
curred in the 1-ha connected fragments, where the num- respectively. Prior to fragmentation, there were five
ber of species in 1995 (two years postfragmentation) records of resident species, including the Black-capped
increased relative to 1993 and 1994 (Friedman’s x2 5 Chickadee, Brown Creeper, and Red-breasted Nut-
4.67, P 5 0.097, df 5 2). These small, connected frag- hatch, in the three 10-ha isolates; however, there were
ments also contained more species in 1995 than did none two years after fragmentation. When only prob-
equally sized isolated fragments and controls (F 5 able breeders were considered, there was no significant
4.047; df 5 2, 6; P 5 0.077). However, the number of difference in species turnover between treatments and
species considered to be probable breeders in the 1-ha controls (F 5 0.52; df 5 2, 30; P 5 0.597); however,
connected fragments did not change significantly ( x2 rates in the 1-ha isolated and connected fragments
5 0.86, P 5 0.649), suggesting that the increase was (94.4%, both cases) were high relative to the controls
due to transient species. (75%), as reflected in the similarity–area curves with
Species diversity.—There was no significant rela- steeper slopes.
tionship between a diversity, as measured by the log Bird abundance.—There was no significant differ-
series index, and area in the controls or treatments prior ence in total abundance of all species in the controls
to fragmentation (1993) (Pearson’s correlation coeffi- or treatments prior to fragmentation (F 5 0.38; df 5
cient: r 5 0.22 to 0.40, P 5 0.203 to 0.557; Fig. 4A), 2, 216; P 5 0.683; Fig. 6A). More individuals were
suggesting that the observed species–area relationships recorded in the isolated fragments than in either the
were an artifact of sample size (see Discussion). There controls or connected fragments one year after frag-
remained no significant relationship in either the con- mentation (F 5 8.43; df 5 2, 216; P , 0.001), but
trols or connected fragments for the two years after differences were not significant two years after frag-
fragmentation (r 5 20.04 to 0.34, P 5 0.365 to 0.950). mentation (F 5 1.57; df 5 2, 216; P 5 0.207; Fig.
However, the relationship was significant in both years 6A), supporting our third prediction concerning tem-
following fragmentation for the isolated fragments (re- porary crowding effects in the recently fragmented ar-
gression: F 5 12.99; df 5 1, 10; P1994 5 0.005, R2 5 eas. Abundance trends did, however, differ among mi-
0.565; F 5 12.80; df 5 1, 10; P1995 5 0.005, R2 5 gratory groups.
0.561; Fig. 4B, C), due primarily to a decrease in di- Prior to fragmentation, numbers of Neotropical mi-
versity in the 1-ha and 10-ha isolates, revealing the grants (NTM) and short-distance (SD) migrants did not
occurrence of a true species–area relationship in these differ among treatments and controls (FNTM 5 1.86; df
sites. 5 2, 216; PNTM 5 0.156, Fig. 6B; FSD 5 0.61; df 5 2,
Species turnover.—We found significant differences 216; PSD 5 0.542, Fig. 6C), although resident (RES)
in the composition of species supported by sites before numbers were lower in connected fragments (F 5 2.49;
and after fragmentation (F 5 2.94; df 5 2, 30; P 5 df 5 2, 216; P 5 0.087, Fig. 6D). Neotropical migrant
0.069; Fig. 5). Isolated fragments were less similar numbers increased in the isolated fragments relative to
overall (they had higher species turnover) than either the controls, but not in the connected fragments, im-
control sites (F 5 4.11; df 5 1, 22; P 5 0.055) or mediately following fragmentation (F 5 5.95; df 5 2,
connected fragments (F 5 5.68; df 5 1, 19; P 5 0.029), 216; P 5 0.003, Bonferroni correction). Two years after
which did not differ in their similarity (F 5 0.01; df fragmentation, however, both isolated and connected
FIG. 4. Log series a diversity of forest birds in treatment

FIG. 3. Species–area curves for forest birds (‘‘all spe- and control areas (A) before fragmentation, (B) one year after
cies’’) in treatment and control areas (A) before fragmenta- fragmentation, and (C) two years after fragmentation. The
tion, (B) one year after fragmentation, and (C) two years after only significant regressions were for the isolated fragments
fragmentation (n 5 12 for controls and isolated fragments in (B) and (C) (n 5 12 for controls and isolated fragments
and n 5 9 for connected fragments). and n 5 9 for connected fragments).
fragments contained fewer Neotropical migrants than in the connected fragments one year following frag-
did control sites (F 5 6.50; df 5 2, 216; P 5 0.002, mentation (F 5 3.52; df 5 2, 216; P 5 0.031, Bonferroni
Bonferroni correction; Fig. 6B). Numbers of short-dis- correction). In the second year, resident numbers were
tance migrants were significantly higher in the isolated lower in both the isolated and connected fragments than
fragments than in connected fragments, but not higher in the controls, and isolated and connected fragments
than in controls, immediately after fragmentation (F 5 were no longer significantly different (F 5 15.68; df 5
6.25; df 5 2, 216; P 5 0.002, Bonferroni correction). 2, 216; P , 0.001, Bonferroni correction; Fig. 6D).
They did not differ among treatments and controls in We had sufficient observations (more than five in
the second year (F 5 2.04; df 5 2, 216; P 5 0.130; any year in both treatments and controls) to analyze
Fig. 6C). Abundance of resident species remained lowest data from 37 species in isolated fragments and 30 spe-
fect on seven species in the isolated fragments and six

species in the connected fragments, and a significant
positive effect on three species in the isolated and two
species in the connected fragments (Table 7). Five of
the seven species that declined due to fragmentation in
the isolated fragments were Neotropical migrants and
the remaining two species were residents, a signifi-
cantly nonrandom distribution with respect to overall
community composition (G 5 6.40, P 5 0.041). All
of the species negatively affected in the connected frag-
ments were Neotropical migrants, also significantly
nonrandom (G 5 4.86, P 5 0.088). There was no sig-
nificant pattern with respect to migratory habit in the
species positively affected by fragmentation in either
the isolated (GI 5 1.98, PI 5 0.372) or connected frag-
FIG. 5. Similarity of bird species assemblages within
treatment and control areas from prefragmentation (1993) to ments (GC 5 0.82, PC 5 0.664).
two years after fragmentation (1995) (n 5 12 for controls and When only trends from before fragmentation to one
isolated fragments and n 5 9 for connected fragments). year after fragmentation were considered (1993–1994),
many of the species that declined significantly over the
longer term in the isolated fragments exhibited increas-
cies in connected fragments, comparing changes in in- es, consistent with our crowding hypothesis (Table 6).
dividual species abundances over the 3-yr course of These species included the Black-throated Green War-
this study (1993–1995) relative to changes observed in bler, Chipping Sparrow, Rose-breasted Grosbeak, and
our controls (Table 6). Measured two years after forest Ruby-crowned Kinglet. This pattern was not apparent
harvesting, fragmentation had a significant negative ef- in the connected fragments.
FIG. 6. Changes in mean (11 SE ) relative abundance per sampling station within and between treatment and control areas
for (A) all bird species, (B) Neotropical migrants, (C) short-distance migrants, and (D) resident species (n 5 93 for controls
and isolated fragments and n 5 33 for connected fragments).
TABLE 6. Total relative abundances of bird species in both treatments and controls before fragmentation (1993), and in the
two years following fragmentation (1994 and 1995). Abundance weightings are in Table 1. Species are listed from most
to least abundant within migration categories (n 5 93 sampling stations for controls and isolated fragments; n 5 33 for
connected fragments).
Controls Isolated fragments Connected fragments

Species 1993 1994 1995 1993 1994 1995 1993 1994 1995
Neotropical migrants
Ovenbird 439.0 247.0 311.0 296.5 208.5 141.0 156.0 99.5 63.5
Red-eyed Vireo 221.5 177.0 236.0 165.0 194.5 173.0 65.0 89.0 97.5
Least Flycatcher 161.5 166.5 167.5 222.5 248.0 253.5 62.5 53.5 75.5
Black-throated Green Warbler 220.0 147.5 167.5 230.5 209.5 142.5 47.0 46.0 28.5
Mourning Warbler 141.0 142.5 144.0 162.0 209.0 210.0 39.0 54.0 43.0
Tennessee Warbler 17.5 35.0 294.0 37.0 43.0 224.0 4.0 6.0 50.0
American Redstart 91.5 79.0 74.5 146.5 146.0 120.5 20.5 20.5 7.0
Connecticut Warbler 78.0 73.0 74.0 113.0 131.0 99.5 47.0 37.0 50.0
Yellow Warbler 59.5 60.5 82.5 91.5 109.0 105.5 33.0 38.0 19.0
Chipping Sparrow 85.0 65.0 131.5 58.0 68.0 52.0 29.0 21.0 33.0
Swainson’s Thrush 64.0 51.5 92.0 41.0 52.0 72.5 33.0 20.5 23.5
Western Tanager 81.0 47.0 48.5 53.0 68.0 42.0 32.5 32.5 26.5
Rose-breasted Grosbeak 34.0 26.5 74.5 65.0 66.0 58.5 33.5 16.0 13.0
Warbling Vireo 42.0 20.0 19.0 55.0 36.0 37.0 11.0 4.0 4.0
Solitary Vireo 46.0 35.0 20.0 29.5 24.0 21.0 15.0 12.0 9.0
Lincoln’s Sparrow 26.5 15.0 7.0 27.0 15.0 7.0 6.0 3.0 0
Ruby-crowned Kinglet 16.0 3.0 22.0 17.0 9.0 7.0 18.0 2.0 11.0
Philadelphia Vireo 20.0 5.0 12.0 26.0 7.0 11.0 10.0 2.0 5.0
Magnolia Warbler 11.0 7.0 11.0 15.0 8.0 4.0
Western Wood Peewee 13.5 12.0 4.0 1.0 0 6.0 2.0 1.0 7.0
Canada Warbler 0 0 5.0 1.0 1.0 5.5
Short-distance migrants
White-throated Sparrow 587.5 453.0 467.5 656.5 602.0 556.0 157.0 118.0 118.0
Yellow-rumped Warbler 360.0 311.5 365.5 302.5 338.0 368.0 145.0 89.5 140.5
Yellow-bellied Sapsucker 116.5 54.0 44.0 82.5 48.0 48.5 46.0 34.0 14.5
Winter Wren 46.0 32.0 75.0 38.0 50.0 59.0 19.0 15.0 33.0
Brown-headed Cowbird 29.5 17.0 15.5 15.0 25.5 12.0 18.5 15.5 12.5
American Robin 4.0 28.5 15.0 0 19.0 13.0 4.0 14.5 7.5
Dark-eyed Junco 17.0 18.0 10.0 20.5 11.5 11.0
Pine Siskin 27.5 3.0 2.0 31.5 2.5 0 11.0 2.0 3.0
Hermit Thrush 7.5 10.0 11.0 8.0 7.0 24.0
Cedar Waxwing 5.0 2.0 9.0 8.5 10.0 18.5
White-winged Crossbill 20.0 1.0 0 24.5 1.0 0.5 6.0 0 0
House Wren 14.0 1.0 1.0 9.0 8.0 12.0
Purple Finch 0 0 7.0 2.0 3.0 5.0
Residents
Red-breasted Nuthatch 11.0 100.5 27.5 15.0 106.5 6.5 3.0 23.0 5.5
Brown Creeper 27.5 52.5 35.0 12.5 41.0 14.0 3.0 10.0 1.0
Black-capped Chickadee 22.0 24.0 29.5 37.0 20.0 21.0 6.0 5.0 1.0
Note: Only species with more than five detections in any year in both treatments and controls are included.
DISCUSSION observed elsewhere. We observed no net loss of species

We made four predictions concerning broad-scale re- from fragmented areas, although the log series a re-
sponses of the avian community to experimental forest vealed changes in the species–area relationship of the
fragmentation: (1) that species loss from fragmented isolated fragments when the influence of sample size
areas would result in species–area relationships with was removed. We documented higher turnover rates
steeper slopes and lower intercepts; (2) that small frag- and crowding effects immediately after fragmentation,
ments would experience high species turnover and as well as significant declines in the abundance of cer-
would lose specialist species of older forest; (3) that tain species in the second year following fragmenta-
abundances of some species would temporarily in- tion, indicating that overall community structure had
crease in recently created forest fragments, due to dis- been altered. We also found some evidence that con-
placement of individuals from adjacent cut areas; and nections between fragments helped to offset these ef-
(4) that connected fragments would maintain their pre- fects.
fragmentation community structure better than would
Species loss due to fragmentation
completely isolated fragments. Our results were con-
sistent with predictions 2–4, but the magnitudes of the We found a significant relationship between number
effects we observed were small compared with those of species and area, and turnover of species in the com-
TABLE 7. Effect of treatment on bird species abundances in isolated and connected forest fragments one and two years after
fragmentation. Species listings are as in Table 6.
Isolated fragments Connected fragments

Year 1 Year 2 Year 1 Year 2
Species Status† Effect‡ P§ Effect P Effect P Effect P
Ovenbird NTM 1 0.007 NS NS 2 0.010
Red-eyed Vireo NTM 1 0.001 NS 1 0.001 1 0.039
Black-throated Green Warbler NTM 1 0.098 2 0.092 1 0.072 NS
Mourning Warbler NTM 1 0.064 NS NS NS
Tennessee Warbler NTM NS 2 ,0.001 NS 2 0.023
Yellow Warbler NTM NS NS NS 2 0.023
Chipping Sparrow NTM NS 2 0.021 NS NS
Swainson’s Thrush NTM NS NS NS 2 0.061
Western Tanager NTM 1 (0.008) NS NS NS
Rose-breasted Grosbeak NTM NS 2 (0.020) 2 0.019 2 ,0.001
Ruby-crowned Kinglet NTM NS 2 0.071 NS 2 0.088
Magnolia Warbler NTM NS NS 1 0.056 NS
Western Wood Peewee NTM NS 2 (0.009) NS 1 0.038
White-throated Sparrow SD 1 0.053 NS NS NS

Yellow-rumped Warbler SD 1 0.015 NS 2 (0.008) NS
Winter Wren SD 1 (0.058) NS NS NS
Brown-headed Cowbird SD 1 0.012 NS NS NS
Hermit Thrush SD NS 1 0.090
House Wren SD 1 (0.012) 1 (0.087) NS NS
Red-breasted Nuthatch RES NS 2 (0.004) 2 0.022 NS

Black-capped Chickadee RES NS 2 0.088 NS NS
† Neotropical migrants (NTM), permanent residents (RES), and short-distance migrants (SD) (after Godfrey 1986, Anon-
ymous 1991).
‡ Change in mean abundance relative to that observed in the controls.
§ All P values are based on t tests, except those in parentheses, which are from Mann-Whitney tests; NS, nonsignificant.
munity through local extinctions and recolonizations, imental study, although more rigorous than retrospec-
but no decrease in species richness after fragmentation. tive approaches, is short-term. However, we might ex-
Contrary to our prediction of steeper slopes and lower pect major responses to our manipulation to occur now
intercepts, we detected no significant change in the rather than later, as regrowth in the cut blocks sur-
species–area relationships either within, or between, rounding the fragments is rapid (.1.5 m in two years),
treatment and control sites in response to experimental potentially ameliorating negative effects (see also
fragmentation. The only significant change in species Stouffer and Bierregaard 1995). In designing this ex-
richness in any size class occurred in the 1-ha con- periment, we estimated that we would be able to detect
nected fragments, where the mean in 1995 (two years a 20% loss of species from a 50-ha area (Schmiegelow
postfragmentation) was 10.0 species, a substantial in- and Hannon 1993), which should exceed the level of
crease from 4.7 species in 1994 (one year postfrag- variation naturally present in the system. Mean annual
mentation) and 6.0 species in 1993 (prefragmentation), variation in species number in a 50-ha control area over
and significantly higher than the mean of 4.3 species the 3-yr course of this study (extrapolated from Table
in 1-ha isolated fragments, and 5.0 species in 1-ha con- 3) was 10.9%, reflecting an estimated loss of three
trols, in the same year. However, numbers of probable species from 1993–1994, and a recovery of these three
breeders remained constant at roughly three species in species from 1994–1995. Therefore, an effect size of
each year. We propose that the proximity of the sam- 20%, roughly twice the variability inherent in the sys-
pling stations in the 1-ha connected fragments to the tem, should have been readily detected, particularly
adjacent, 100 m wide riparian buffer strip increased the since the estimates of variance we used for our original
probability that species moving through or breeding in power analyses are higher than those we observed.
those buffers were incidentally sampled, relative to ei- We attempted to further refine our ability to detect
ther the 10-ha or 40-ha connected fragments, or any of meaningful changes in species richness by defining a
the isolated fragments. subset of species that we classed as probable breeders.
Community collapse or relaxation following habitat However, this approach was limited by our reliance on
fragmentation (Brown 1971), as reflected by changes point count data and by problems associated with sam-
in species–area relationships, is usually measured over pling rare species. Although these limitations were con-
the course of decades or longer, and most cases of sistent across treatments and years and, thus, did not
species loss from forest fragments are from areas where bias our results, some species may not have been sam-
regional deforestation occurred long ago. This exper- pled adequately, due either to breeding phenology or
to detectability, given our criteria for defining probable fragments relative to the controls, suggesting a
breeders (see Table 1). ‘‘crowding’’ effect. We detected no differences in the
overall abundance of birds between treatments and con-
Patterns in species diversity and turnover trols two years after fragmentation, consistent with our
Changes in diversity.—The absence of a relationship third prediction that increases in abundance would be
between bird diversity and area in both treatments and temporary.
controls before fragmentation is not surprising, given When we examined the structure of the communities
our experimental design and the use of the log series more closely, we discovered that abundance responses
a index. We attempted to minimize habitat variation varied with respect to migratory strategy of the birds.
through careful selection of our study areas, and the a Overall, Neotropical migrants exhibited the strongest
index removes the effect of sample size, so the two crowding effect immediately after fragmentation,
factors that influence species–area relationships were short-distance migrants exhibited an intermediate level
standardized. Hence, the occurrence of a significant of crowding, and numbers of residents remained com-
diversity–area relationship (and true species–area re- parable among treatments and controls. These differ-
lationship) in the isolated fragments following frag- ences may be related, in part, to breeding phenology.
mentation, but not in other sites, is particularly inter- Harvesting was completed in February; therefore, res-
esting. The relationship resulted primarily from a re- ident birds had a longer time to respond to the exper-
duction in diversity in the 1-ha and 10-ha isolates, but imental fragmentation by prospecting for new territo-
there was no consistent pattern with respect to either ries prior to breeding. However, the overall low den-
changes in numbers of individuals or changes in num- sities of resident species might also have made crowd-
bers of species across years or among replicates. Hab- ing difficult to detect. Returning migrants were more
itat fragmentation created diversity relationships in the constrained in their ability to respond to the habitat
isolated fragments consistent with predictions from is- alteration, Neotropical migrants most so, given the nar-
land biogeography theory, and it is likely that changes row window of time in which territory establishment,
in both population sizes and habitat diversity were im- nest-building, laying, incubation, and rearing must take
portant factors. Whether these will lead to longer term place. This effect would decline over time, because
species loss is not known. adult mortality would reduce philopatry in subsequent
Changes in species composition.—The significant seasons, and juveniles might prospect for future ter-
positive relationship that we observed between com- ritories at the end of the breeding season, prior to mi-
munity similarity over time and area is consistent with gration (Brewer and Harrison 1975, Adams and Brewer
results from other studies (e.g., Hinsley et al. 1995), 1981, Morton et al. 1991). Hagan et al. (1996), based
and suggests that community stability increases with on results from Bierregaard et al. (1992) for resident
size. Over the course of this study, the communities in birds in tropical forests, predicted that it could take two
isolated fragments experienced slightly, but signifi- or three breeding seasons (years) for elevated Neo-
cantly, higher species turnover than in either connected tropical migrant numbers to return to prefragmentation
fragments or control sites, with the effect being most levels, or for negative numerical effects to develop, in
pronounced in smaller isolates, consistent with our sec- temperate regions.
ond prediction. Since there was no significant loss of Two years following fragmentation, Neotropical mi-
species from these areas over the same time period grants and residents were less abundant in both the
(1993–1995), there was substantial replacement of spe- isolated and connected fragments than in the controls,
cies present in the prefragmentation community. The but there was no difference among short-distance mi-
resident species we studied, all of which prefer older grants. This pattern also held at the species level: all
forest (Schieck and Nietfeld 1995), most frequently 10 bird species that exhibited significant negative re-
disappeared from fragmented areas. However, the pat- sponses to fragmentation in the isolated or connected
tern of replacement by other species was not consistent, sites were either Neotropical migrants or residents, a
supporting the notion of community instability that was nonrandom result with respect to community compo-
also suggested by changes in diversity. sition. Species preferring older forest accounted for
seven of the 10 species that declined in either the iso-
Changes in abundance lated or connected fragments, whereas the four species
Several studies have documented crowding of birds that increased in abundance following fragmentation
in habitat fragments immediately postfragmentation, either prefer younger forest or are not typically forest
followed by relaxation of the community in subsequent species (see Semenchuck 1992, Schieck and Nietfeld
years (Whitcomb et al. 1981, Bierregaard and Lovejoy 1995). Several additional species, also Neotropical mi-
1989, Darveau et al. 1995). We found no absolute in- grant, old-forest specialists, showed nonsignificant
crease in overall abundance of birds in the isolated downward trends in abundance.
fragments immediately postfragmentation. However, These results are generally consistent with studies
although bird abundance decreased in the controls in from both eastern deciduous forests (Ambuel and Tem-
1994, there was a significant increase in the isolated ple 1983, Lynch and Whigham 1984) and the eastern
boreal forest (Darveau et al. 1995), and with the wide- gimes that have created such mosaics, have led to sug-
spread population declines in forest-breeding Neotrop- gestions that bird species occupying these habitats may
ical migrants reported from more eastern regions of be preadapted to large-scale disturbance and fragmen-
their breeding range (e.g., Robbins et al. 1989, Askins tation (e.g., Welsh 1987). Support for this comes pri-
1995). Ambuel and Temple (1983) suggested that some marily from work in western Palaearctic forests (see
forest-dwelling, long-distance migrants were actually Haila 1994), where the numerically dominant species
excluded from forest fragments by forest edge and tend to be habitat generalists (Haila et al. 1994, Hans-
farmland species (most of which were short-distance son 1994). However, Neotropical migrants, most of
migrants or residents), through competition, predation, which are habitat specialists, predominate in the Ne-
or nest parasitism. Darveau et al. (1995) report a de- arctic (Mönkkönen 1994, Mönkkönen and Welsh
crease in forest-dwelling species, and an increase in 1994). Such specialization might compromise their
ubiquitous species in riparian buffer strips created by ability to adjust to the rapid landscape changes brought
clear-cutting of adjacent areas. From Table 1 of their about by industrial forestry. Secondly, there is a long
study, we classified species in these groups by their history of human-induced habitat change in western
migratory habit: 71.4% of all territories of forest-dwell- Palaearctic forests, leading Angelstam (1992) to sug-
ing species belonged to Neotropical migrants, and gest that many of the species sensitive to forest frag-
94.3% of the territories of ubiquitous species were held mentation became extinct hundreds of years ago, thus
by short-distance migrants. In all cases, response of confounding conclusions based on contemporary pat-
bird species to fragmentation differed with respect to terns of response. The bird communities of the western
migratory habit. Nearctic boreal forest have not been exposed to such
Connectivity and fragmentation effects human-induced landscape alterations, and might con-
tain more sensitive species.
Our results indicate that the connected fragments Nevertheless, some studies from Palaearctic boreal
were less affected by fragmentation than were the iso- forests have identified groups of birds sensitive to frag-
lated areas, consistent with our fourth prediction, but mentation and loss of older forest (e.g., Helle and Jär-
overall differences were small. The larger effective vinen 1986, Väisänen et al. 1986). In particular, wood-
sizes of the connected fragments may, alone, have been peckers and other hole-nesters (Angelstam and Miku-
enough to lessen the effects seen in the smaller, isolated siński 1994, Virkkala et al. 1994) and several additional
fragments, since the habitat in the adjacent buffer strips northern taiga species (Virkkala 1987) are of concern
increased the relative area of the 1-ha and 10-ha con- due to declining numbers. Many of these species have
nected sites by 100% and 40%, respectively. Also, the Nearctic counterparts in our resident birds (see Haila
availability of adjacent habitat in the buffer strips, and and Järvinen 1990). Furthermore, recent studies in both
the fact that the amount of habitat removed around the Palaearctic and Nearctic boreal forests indicate that
connected fragments (and, consequently, the potential small-scale gap dynamics may play an important role
pool of displaced individuals) was 25% lower than for in structuring the forest at both local and regional scales
the isolated fragments, would dilute any crowding ef- (Kuuluvainen 1994; S. G. Cumming, F. K. A. Schmie-
fect. In addition, concurrent studies in the area indi- gelow, and P. J. Burton, unpublished manuscript), ar-
cated that forest birds moved through the adjoining
guing against the uncritical acceptance of a natural dis-
buffer strips much more frequently than across clear-
turbance paradigm based on large, catastrophic events
cuts, and that juvenile dispersers used the buffer strips
such as fire and insect outbreaks.
as corridors (Machtans et al. 1996), which should fur-
ther mitigate any fragmentation effects. The role of the landscape matrix
These results also suggest that distances as small as
Studies of fragmentation conducted in agricultural
200 m can effectively isolate many forest birds during
or suburban landscapes have often concluded that
the breeding season (see also Soulé et al. 1988). Sur-
losses of species, or declines in abundance, are the
veys of clearcuts adjacent to both the connected and
result of increased rates of nest parasitism and preda-
isolated fragments recorded very few movements of
tion by edge-related species (e.g., Brittingham and
birds associated with older forest across such areas
Temple 1983, Wilcove 1985, Martin 1992b). Fragmen-
(Machtans et al. 1996; F. K. A. Schmiegelow, unpub-
tation by agriculture favors cowbirds, since it provides
lished data). Experimental work by A. Desrochers and
both feeding and breeding habitats. Fragmentation by
S. J. Hannon (in press) supports the idea that clearcuts
forest harvesting changes forest age, composition, and
represent behavioral, rather than physiological, barriers
structure, but does not create feeding habitat for cow-
to forest birds, and argues for the consideration of con-
birds. We detected no significant change in numbers of
straints beyond dispersal ability per se.
Brown-headed Cowbirds, an obligate nest parasite, two
Natural and human-induced disturbances in years after fragmentation, and brood parasitism rates
boreal systems in our study area are extremely low (M.-A. Villard and
The spatial and temporal patchiness that character- S. J. Hannon, unpublished data). We also doubt that
izes the boreal forest, and the natural disturbance re- predation will significantly lower nesting success, since
forest harvesting is not creating a matrix that similarly we measured only short-term, broad-scale responses
favors many nest predators. In an experiment in ad- based on relatively coarse community measures. The
jacent forest, Cotterill (1996) found no immediate in- nonsignificant trends we observed could become sig-
crease in overall predation rates on artificial nests as a nificant declines or losses over time. We are continuing
result of habitat fragmentation. to sample the fragments in order to monitor longer term
The influence of regional patterns of forest cover on effects. Furthermore, estimates of abundance may not
the response of birds to fragmentation is being increas- provide a reliable indicator of habitat quality (Van
ingly recognized (e.g., Askins and Philbrick 1987, Horne 1983). Ongoing studies of avian productivity
Freemark and Collins 1992, McGarigal and McComb address this concern (F. K. A. Schmiegelow, M.-A. Vil-
1995, Robinson et al. 1995). When large tracts of forest lard and S. J. Hannon). Second, we did not analyze
exist regionally, populations of birds in smaller, more data from several groups of bird species that might be
isolated patches of forest may be maintained through more sensitive to fragmentation at this scale. For ex-
a ‘‘rescue effect’’ (Brown and Kodrick-Brown 1977). ample, woodpeckers and raptors, many of which are
Presently, our study area exists in a landscape matrix resident species, generally require larger breeding and
dominated by older, mixed forest, which may buffer foraging areas than do the species reported here. We
local impacts of fragmentation. The boreal mixed-wood have expanded our surveys to better sample wood-
mosaic will remain forest dominated, but the relative peckers, and are currently documenting changes in rap-
proportions of stands of different age and composition tor communities over a larger scale (B. Olsen, G. Court,
are likely to change dramatically as harvesting pro- and S. J. Hannon). Third, any effects may have been
ceeds. mitigated by the large areas of old, mixed forest still
Cumming et al. (1994) modelled various boreal present regionally. The availability of such habitat in
mixed-wood management strategies, concluding that, the future will be substantially reduced, due to the
without substantial reductions in harvest levels or in- widespread industrial forestry in progress in the boreal
creases in operating costs, most of the region’s older, mixed-wood forest.
mixed stands will be converted to younger, single-spe-
cies stands. Habitat loss may result in overall popu- ACKNOWLEDGMENTS
lation declines of certain bird species, but this is not a We thank the many field assistants involved in this study,
fragmentation effect per se (see Haila 1986). However, particularly R. Brown, and C. McCallum for field and tech-
area-related edge effects, such as reduction in habitat nical support. The Calling Lake Ranger Station and Meanook
Biological Research Station provided logistical support. Al-
quality due to changes in vegetation (Matlack 1993, berta Pacific Forest Industries and Vanderwell Contracting
Esseen 1994, Pettersson et al. 1995), and competitive conducted the logging necessary for our experimental design,
exclusion or replacement by species able to utilize the and the cooperation and involvement of members of the Al-
surrounding matrix (e.g., Ambuel and Temple 1983, berta Forest Service and Alberta Pacific Forest Industries
have been valuable assets to this project. S. Hejl, J. N. M.
Haila et al. 1989, Ims and Rolstad 1993), are likely to
Smith, K. Sullivan, M.-A. Villard, and an anonymous re-
exacerbate habitat loss and increase the minimum area viewer provided valuable comments on earlier drafts of this
required by old-forest specialists. Important consider- manuscript, and S. G. Cumming provided computer support.
ations include the ability of individuals to successfully Funding for this work was provided by the National Fish and
reproduce in remnant patches of older forest, the avail- Wildlife Foundation, Wildlife Habitat Canada, the J. K. Coo-
per Foundation, the Canadian Circumpolar Institute, the Ca-
ability of such patches over time, and changes in po- nadian Wildlife Service, Forestry Canada’s Canada–Alberta
tential edge effects over time as the surrounding forest Partnership in Forestry Program, Alberta Environmental Pro-
regenerates. As Haila et al. (1993), among others, have tection, Alberta Recreation, Parks and Wildlife, Alberta Pa-
pointed out, a simplified concept of fragmentation is cific Forest Industries, Daishowa-Marubeni Industries, the
not adequate when considering forest patches within National Eco-Research Tri-Council, and the Natural Sciences
and Engineering Research Council of Canada.
dynamic habitat mosaics.
LITERATURE CITED
CONCLUSIONS
Adams, R. J., and R. Brewer. 1981. Autumn selection of
Although we observed some significant negative ef- breeding location by Field Sparrows. Auk 98:629–631.
fects of the experimental fragmentation, consistent with Ambuel, B., and S. A. Temple. 1983. Area-dependent
our predictions, and some interesting trends, magni- changes in the bird communities and vegetation of southern
tudes were small, given the extent of our manipulations. Wisconsin forests. Ecology 64:1057–1068.
Angelstam, P. 1992. Conservation of communities: the im-
Where effects did occur, they were most pronounced portance of edges, surroundings, and landscape mosaic
in the smaller isolates, and Neotropical migrant and structure. Pages 9–70 in L. Hannson, editor. Ecological
resident species exhibited the greatest sensitivity. Re- principles of nature conservation: applications in temperate
sponses were generally smaller or nonexistent in the and boreal environments. Elsevier Science, Barking, Essex,
connected fragments relative to the isolated fragments. UK.
Angelstam, P., and G. Mikusiński. 1994. Woodpecker as-
It appears that the breeding boreal bird community is semblages in natural and managed boreal and hemiboreal
quite resilient to major disturbances. However, there forest, a review. Annales Zoologici Fennici 31:157–172.
are several important caveats to our conclusions. First, Anonymous, 1991. Preliminary lists of migrants for Partners
in Flight Neotropical migratory bird conservation program. in wooded patches on an agricultural landscape. Conser-
Partners in Flight 2:30. vation Biology 9:845–854.
Askins, R. A. 1995. Hostile landscapes and the decline of Hagan, J. M., W. M. Vander Haegen, and P. S. McKinley.
migratory songbirds. Science 267:1956–1957. 1996. The early development of forest fragmentation ef-
Askins, R. A., and M. J. Philbrick. 1987. Effect of changes fects on birds. Conservation Biology 10:188–202.
in regional forest abundance on the decline and recovery Haila, Y. 1986. North European land birds in forest frag-
of a forest bird community. Wilson Bulletin 99:7–21. ments: evidence for area effects? Pages 315–320 in J. Ver-
Barker, R. J., and J. R. Sauer. 1995. Statistical aspects of ner, M. L. Morrison, and C. J. Ralph, editors. Wildlife 2000:
point count sampling. Pages 125–130 in C. J. Ralph, J. R. modeling habitat relationships of terrestrial vertebrates.
Sauer, and S. Droege, editors. Monitoring bird populations University of Wisconsin Press, Madison, Wisconsin, USA.
by point counts. General Technical Report PSW-GTR-149. Haila, Y. 1994. Preserving ecological diversity in boreal for-
Pacific Southwest Research Station, U.S. Forest Service, ests: ecological background, research, and management.
U.S. Department of Agriculture. Annales Zoologici Fennici 31:203–217.
Bierregaard, R. O., and T. E. Lovejoy. 1989. Effects of forest Haila, Y., I. K. Hanski, J. Niemeiä, P. Punttila, S. Raivio, and
fragmentation on Amazonian understory bird communities. H. Tukia. 1994. Forestry and the boreal fauna: matching
Acta Amazonica 19:215–241. management with natural forest dynamics. Annales Zool-
Bierregaard, R. O., T. E. Lovejoy, V. Kapos, A. A. dos Santos, ogici Fennici 31:187–202.
and R. W. Hutchings. 1992. The biological dynamics of Haila, Y., I. K. Hanski, and S. Raivio. 1989. Methodology
tropical rainforest fragments. A prospective comparison of for studying the minimum habitat requirements of forest
fragments and continuous forest. BioScience 42:859–866. birds. Annales Zoologici Fennici 26:173–180.
Brewer, R., and K. G. Harrison. 1975. The time of habitat Haila, Y., and O. Järvinen. 1990. Northern conifer forests
selection by birds. Ibis 117:521–522. and their breeding bird assemblages. Pages 61–85 in A.
Brittingham, M. C., and S. A. Temple. 1983. Have Cowbirds Keast, editor. Biogeography and ecology of forest bird com-
caused forest songbirds to decline? BioScience 33:31–35. munities. SPB Academic Publishing, The Hague, The Neth-
Brown, J. H. 1971. Mammals on mountaintops: nonequilib- erlands.
rium insular biogeography. American Naturalist 105:467– Haila, Y., D. Saunders, and R. Hobbs. 1993. What do we
478. presently know understand about ecosystem fragmenta-
Brown, J. H., and A. Kodric-Brown. 1977. Turnover rate in tion? Pages 45–55 in D. Saunders, R. J. Hobbs, and P.
insular biogeography: effects of immigration on extinction. Ehrlich, editors. Nature conservation 3: reconstruction of
Ecology 58:445–449. fragmented ecosystems. Surrey Beatty and Sons, Chipping
Conover, W. J. 1980. Practical nonparametric statistics. Sec- Norton, New South Wales, Australia.
ond edition. John Wiley, New York, New York, USA. Hansson, L. 1994. Vertebrate distributions relative to clear-
Cotterill, S. E. 1996. Effect of clearcutting on artificial egg cut edges in a boreal forest landscape. Landscape Ecology
predation in boreal mixed-wood forests in north-central Al- 9:105–115.
berta. Thesis. University of Alberta, Edmonton, Alberta, Helle, P. 1984. Effect of habitat area on breeding bird com-
Canada. munities in northeastern Finland. Annales Zoologici Fen-
Cumming, S. G., P. J. Burton, S. Prahacs, and M. R. Garland. nici 21:421–425.
1994. Potential conflicts between timber supply and habitat Helle, P., and O. Järvinen. 1986. Population trends of north
protection in the boreal mixed-wood of Alberta, Canada: Finnish land birds in relation to their habitat selection and
a simulation study. Forest Ecology and Management 68: changes in forest structure. Oikos 46:107–115.
281–302. Hinsley, S. A., P. E. Bellamy, and I. Newton. 1995. Bird
Darveau, M., P. Beauchesne, L. Belanger, J. Huot, and P. species turnover and stochastic extinction in woodland
Larue. 1995. Riparian forest strips as habitat for breeding fragments. Ecography 18:41–50.
birds in the boreal forest. Journal of Wildlife Management Hobbs, R. J. 1992. The role of corridors in conservation:
59:67–78. solution or bandwagon? Trends in Ecology and Evolution
Desrochers, A., and S. J. Hannon. In press. Gap crossing 7:389–392.
decisions by dispersing forest songbirds. Conservation Bi- Ims, R. A., and J. Rolstad. 1993. Predicting space use re-
ology. sponses to habitat fragmentation: can voles Microtus oec-
Esseen, P.-A. 1994. Tree mortality patterns after experimen- onomus serve as an experimental model system (EMS) for
tal fragmentation of an old-growth conifer forest. Biolog- Capercaillie grouse Tetrao urogallus in boreal forest? Bi-
ical Conservation 68:19–28. ological Conservation 63:261–268.
Fisher, R. A., A. S. Corbet, and C. B. Williams. 1943. The Krebs, C. J. 1989. Ecological methodology. Harper and Row,
relation between the number of species and the number of New York, New York, USA.
individuals in a random sample of an animal population. Kuuluvainen, T. 1994. Gap disturbance, ground microto-
Journal of Animal Ecology 12:75–84. pography, and the regeneration dynamics of boreal conif-
Freemark, K., and B. Collins. 1992. Landscape ecology of erous forests in Finland: a review. Annales Zoologici Fen-
birds breeding in temperate forest fragments. Pages 443– nici 31:35–51.
454 in J. Hagan and D. J. Johnston, editors. Conservation Legendre, L., and P. Legendre. 1983. Numerical ecology.
of Neotropical migrants. Smithsonian Institution, Wash- Elsevier Scientific Publishing, New York, New York, USA.
ington, D.C., USA. Levins, R. 1970. Extinction. Pages 77–107 in M. Gersten-
Freemark, K. E., and H. G. Merriam. 1986. Importance of haber, editor. Some mathematical questions in biology. Lec-
area and habitat heterogeneity to bird assemblages in tem- ture on mathematics in the life sciences, Volume 2. Amer-
perate forest fragments. Biological Conservation 36:115– ican Mathematical Society, Providence, Rhode Island,
141. USA.
Gibbs, J. P., and J. Faaborg. 1990. Estimating the viability Lynch, J. F., and D. F. Whigham. 1984. Effects of forest
of Ovenbird and Kentucky Warbler populations in forest fragmentation on breeding bird communities in Maryland,
fragments. Conservation Biology 4:193–196. USA. Biological Conservation 28:287–384.
Godfrey, W. E. 1986. The birds of Canada. Revised edition. MacArthur, R. H., and E. O. Wilson. 1963. An equilibrium
National Museums of Canada, Ottawa, Ontario Canada. theory of insular zoogeography. Evolution 17:373–387.
Haas, C. A. 1995. Dispersal and use of corridors by birds MacArthur, R. H., and E. O. Wilson. 1967. The theory of
island biogeography. Princeton University Press, Princeton, United States Department of the Interior, Fish and Wildlife
New Jersey, USA. Service Resource Publication 157. Washington, D.C., USA.
Machtans, C. S., M.-.A. Villard, and S. J. Hannon. 1996. Use Robbins, C. S., J. R. Sauer, R. S. Greenberg, and S. Droege.
of riparian buffer strips as movement corridors by forest 1989. Population declines in North American birds that
birds. Conservation Biology 10:1366–1379. migrate to the neotropics. Proceedings of the National
Magurran, A. E. 1988. Ecological diversity and its mea- Academy of Sciences (USA) 86:7658–7662.
surement. Princeton University Press, Princeton, New Jer- Robinson, S. K., F. R. Thompson, T. M. Donovan, D. R.
sey, USA. Whitehead, and J. Faaborg. 1995. Regional forest frag-
Mann, C. C., and M. L. Plummer. 1993. The high cost of mentation and the nesting success of migratory birds. Sci-
biodiversity. Science 260:1868–1871. ence 267:1987–1990.
Martin, T. E. 1992a. Breeding Biology Research database Rosenzweig, M. L. 1995. Species diversity in space and time.
(BBIRD) program: general overview and standardized pro- Cambridge University Press, Cambridge, UK.
tocols. U.S. Department of the Interior, Fish and Wildlife Rowe, J. S. 1972. Forest regions of Canada. Canadian Forest
Service, Arkansas Cooperative Fish and Wildlife Research Service Publication Number 1300.
Unit, University of Arkansas, Fayetteville, Arkansas, USA. Schieck, J., and M. Nietfeld. 1995. Bird species richness and
. 1992b. Breeding productivity considerations: what abundance in relation to stand age and structure in aspen
are the appropriate habitat features for management? Pages mixed-wood forests in Alberta. Pages 115–158 in J. B.
455–473 in J. M. Hagan and D. W. Johnston, editors. Con- Stelfox, editor. Relationships between stand age, stand
servation of Neotropical migrant landbirds. Smithsonian structure, and biodiversity in aspen mixed-wood forest in
Institution, Washington, D.C., USA. Alberta. Jointly published by the Alberta Environmental
Matlack, G. R. 1993. Microenvironment variation within and Centre, Vegreville, Alberta, and the Canadian Forest Ser-
among forest edge sites in the eastern United States. Bi- vice, Edmonton, Alberta, Canada.
ological Conservation 66:185–194. Schmiegelow, F. K. A. 1992. The use of atlas data to test
Maxwell, S. E., and H. D. Delaney. 1990. Designing ex- appropriate hypotheses about faunal collapse. Pages 67–74
periments and analyzing data: a model comparison per- in G. B. Ingram and M. R. Moss, editors. Landscape ap-
spective. Wadsworth Publishing, Belmont, California, proaches to wildlife and ecosystem management in Canada.
USA. Polyscience, Montreal, Ontario, Canada.
May, R. M. 1975. Patterns of species abundance and diver- Schmiegelow, F. K. A., and S. J. Hannon. 1993. Adaptive
sity. Pages 81–120 in M. L. Cody and J. M. Diamond, management, adaptive science, and the effects of forest
editors. Ecology and evolution of communities. Belknap fragmentation on boreal birds in northern Alberta. Trans-
Press, Harvard University, Cambridge, Massachusetts, actions of the 58th North American Wildlife and Natural
USA. Resources Conference: 584–598.
McGarigal, K., and W. C. McComb. 1995. Relationships Semenchuck, G. P., editor. 1992. The atlas of breeding birds
between landscape structure and breeding birds in the Or- of Alberta. Federation of Alberta Naturalists, Edmonton,
egon coast range. Ecological Monographs 65:235–260. Alberta, Canada.
Mönkkönen, M. 1994. Diversity patterns in Palaearctic and Simberloff, D. 1994. How forest fragmentation hurts species
Nearctic forest bird assemblages. Journal of Biogeography and what to do about it. Pages 85–90 in W. W. Covington
21:183–195. and L. F. DeBano, editors. Sustainable ecological systems:
implementing an ecological approach to land management.
Mönkkönen, M., and D. A. Welsh. 1994. A biogeographical
U.S. Department of Agriculture, U.S. Forest Service Gen-
hypothesis on the effects of human-caused landscape
eral Technical Report RM-247.
changes on the forest bird communities of Europe and
Simberloff, D., J. A. Farr, J. Cox, and D. W. Mehlman. 1992.
North America. Annales Zoologici Fennici 31:61–70.
Movement corridors: conservation bargains or poor in-
Morton, M. L., M. W. Wakamatsu, M. E. Pereyra, and G. A.
vestments? Conservation Biology 6:493–504.
Morton. 1991. Postfledging dispersal, habitat imprinting, Smith, S. M. 1995. Distribution-free and robust statistical
and philopatry in a montane, migratory sparrow. Ornis methods: viable alternatives to parametric statistics? Ecol-
Scandinavica 22:98–106. ogy 76:1997–1998.
Norusis, M. J. 1993. SPSS for Windows: base system user’s Soulé, M. E., D. T. Bolger, and A. C. Alberts. 1988. Re-
guide release 6.0. SPSS, Chicago, Illinois, USA. constructed dynamics of extinctions of chapparal-requiring
Opdam, R., G. Rijsdijk, and F. Hustings. 1985. Bird com- birds in urban habitat islands. Conservation Biology 2:75–
munities in small woods in an agricultural landscape: ef- 92.
fects of area and isolation. Biological Conservation 34: Southwood, T. R. E. 1978. Ecological methods. Chapman
333–352. and Hall, London, UK.
Peterman, R. M. 1990. Statistical power analysis can im- Stelfox, J. B., editor. 1995. Relationships between stand age,
prove fisheries research and management. Canadian Journal stand structure, and biodiversity in aspen mixed-wood for-
of Fisheries and Aquatic Sciences 47:2–15. est in Alberta. Jointly published by Alberta Environmental
Petterson, R. B., J. P. Ball, K.-E. Renhorn, P.-A. Esseen, and Centre, Vegreville, Alberta, and the Canadian Forest Ser-
K. Sjoberg. 1995. Invertebrate communities in boreal for- vice, Edmonton, Alberta, Canada.
est canopies, as influenced by forestry and lichens, with Stouffer, P. C., and R. O. Bierregaard, Jr. 1995. Use of Am-
implications for passerine birds. Biological Conservation azonian forest fragments by understory insectivorous birds.
74:57–63. Ecology 76:2429–2445.
Preston, F. W. 1948. The commonness, and rarity, of species. Strong, W. L., and K. R. Leggat. 1981. Ecoregions of Al-
Ecology 29:254–283. berta. Alberta Energy and Natural Resources Technical Re-
Ralph, C. J., G. R. Geupel, P. Pyle, T. E. Martin, and D. F. port T/4. Edmonton, Alberta, Canada.
DeSante. 1993. Handbook of field methods for monitoring Taylor, L. R. 1978. Bates, Williams, Hutchinson: a variety
landbirds. General Technical Report PSW-GTR-144. Pa- of diversities. Pages 1–18 in L. A. Mound and N. Warloff,
cific Southwest Research Station, U.S. Forest Service, U.S. editors. Diversity of insect faunas. Ninth Symposium of
Department of Agriculture. the Royal Entomological Society. Blackwell Scientific, Ox-
Robbins, C. S., D. Bystrak, and P. H. Geissler. 1986. The ford, UK.
breeding bird survey: its first fifteen years, 1965–1979. Taylor, L. R., R. A. Kempton, and I. P. Woiwood. 1976.
Diversity statistics and the log-series model. Journal of Welsh, D. A. 1987. The influence of forest harvesting on
Animal Ecology 45:255–272. mixed deciduous–coniferous boreal bird communities in
Väisänen, R. A., O. Järvinen, and P. Rauhala. 1986. How Ontario. Acta Oecologia 8:247–252.
are extensive, human-caused habitat alterations expressed Welsh, D. A., and S. C. Lougheed. 1996. Relationships of
on the scale of local bird populations in boreal forests? bird community structure and species distributions to two
Ornis Scandinavica 17:282–292. environmental gradients in the northern boreal forest. Ecog-
Van Horne, B. 1983. Density as a misleading indicator of raphy 19:194–208.
habitat quality. Journal of Wildlife Management 47:893– Whitcomb, R. F., C. S. Robbins, J. F. Lynch, B. L. Whitcomb,
901. M. K. Klimliewcz, and D. Bystrak. 1981. Effects of forest
Villard, M.-A., P. R. Martin, and C. G. Drummond. 1993. fragmentation on the avifauna of the eastern deciduous for-
Habitat fragmentation and pairing success in the Ovenbird est. Pages 125–205 in R. C. Burgess and D. M. Sharpe,
(Seiurus aurocapillus). Auk 110:759–768. editors. Forest island dynamics in man-dominated land-
scapes. Springer-Verlag, New York, New York, USA.
Virkkala, R. 1987. Effects of forest management on birds
Whittaker, R. H. 1965. Dominance and diversity in land plant
breeding in northern Finland. Annales Zoologici Fennici communities. Science 147:250–260.
24:281–294. Wilcove, D. S. 1985. Nest predation in forest tracts and the
Virkkala, R., A. Rajasärkkä, R. A. Väisänen, M. Vickholm, decline of migratory song birds. Ecology 66:1211–1214.
and E. Virolainen. 1994. Conservation value of nature Williams, C. B. 1964. Patterns in the balance of nature and
reserves: do hole-nesting birds prefer protected forests in related problems in quantitative ecology. Academic Press,
southern Finland? Annales Zoologici Fennici 31:173–186. London, UK.
von Ende, C. N. 1993. Repeated-measures analysis: growth Winer, B. J., D. R. Brown, and K. M. Michels. 1991. Sta-
and other time-dependent measures. Pages 113–137 in S. tistical principles in experimental design. Third edition.
M. Scheiner and J. Gurevitch, editors. Design and analysis McGraw-Hill, New York, New York, USA.
of ecological experiments. Chapman and Hall, New York, Wolda, B. J. 1981. Similarity indices, sample size, and di-
New York, USA. versity. Oecologia 50:296–302.
APPENDIX
Abundance of bird species in older, aspen-dominated forest, at Calling Lake, Alberta, Canada, from 1993 to 1995. Species
are listed from most to least abundant.
Common name Scientific name Status† Abundance‡ Detections§

White-throated Sparrow Zonotrichia albicolis SD 3715.5 3856
Yellow-rumped Warbler Dendroica coronata SD 2420.5 2452
Ovenbird Seiurus aurocapillus NTM 1962.0 1974
Red-eyed Vireo Vireo olivaceus NTM 1418.5 1427
Least Flycatcher Empidonax minimus NTM 1411.0 1441
Black-throated Green Warbler Dendroica virens NTM 1239.0 1241
Mourning Warbler Oporornis philadelphia NTM 1144.5 1146
Tennessee Warbler Vermivora peregrina NTM 710.5 708
American Redstart Setophaga ruticilla NTM 706.0 710
Connecticut Warbler Oporornis agilis NTM 702.5 703
Yellow Warbler Dendroica petechia NTM 598.5 601
Chipping Sparrow Spizella passerina NTM 542.5 545
Yellow-bellied Sapsucker Sphyrapicus varius SD 488.0 618
Swainson’s Thrush Catharus ustulatus NTM 450.0 500
Western Tanager Piranga ludoviciana NTM 431.0 460
Rose-breasted Grosbeak Pheucticus ludoviciana NTM 387.0 393
Winter Wren Troglodytes troglodytes SD 367.0 369
Red-breasted Nuthatch Sitta canadensis RES 298.5 308
Warbling Vireo Vireo gilvus NTM 228.0 229
Solitary Vireo Vireo solitarius NTM 211.5 212
Brown Creeper Certhia americana RES 196.5 208
Black-capped Chickadee Parus atricapillus RES 165.5 194
Brown-headed Cowbird Molothrus ater SD 160.0 186
Lincoln’s Sparrow Melospiza lincolnii NTM 106.5 107
American Robin Turdus migratorius SD 105.5 119
Ruby-crowned Kinglet Regulus calendula SD 105.0 105
Philadelphia Vireo Vireo philadelphicus NTM 98.0 98
Dark-eyed Junco Junco hyemalis SD 96.5 98
Pine Siskin Carduelis pinus SD 82.5 91
Hermit Thrush Catharus guttatus SD 78.5 84
Magnolia Warbler Dendroica magnolia NTM 60.0 60
Cedar Waxwing Bombycilla cedrorum SD 55.5 65
White-winged Crossbill Loxia leucoptera SD 53.0 56
Western Wood Peewee Contopus sordidulus NTM 46.5 47
Black and White Warbler Mniotilta varia NTM 44.5 46
Golden-crowned Kinglet Regulus satrapa SD 42.5 48
House Wren Troglodytes aedon SD 37.0 37
Blackpoll Warbler Dendroica striata NTM 26.5 27
Purple Finch Carpodacus purpureus SD 18.5 19
Boreal Chickadee Parus hudsonicus RES 16.0 22
Orange-crowned Warbler Vermivora celeta NTM 14.0 14
Canada Warbler Wilsonia canadensis NTM 13.5 14
Common Yellowthroat Geothlypis trichas NTM 8.0 8
Alder Flycatcher Empidonax alnorum NTM 8.0 9
Northern Oriole Icterus galbula NTM 8.0 8
Bay-breasted Warbler Dendroica castanea NTM 6.0 6
White-breasted Nuthatch Sitta carolinensis RES 6.0 6
Eastern Phoebe Sayornis phoebe SD 5.0 5
Olive-sided Flycatcher Contopus borealis NTM 5.0 5
Evening Grosbeak Coccothraustes vespertenus SD 3.5 4
Palm Warbler Dendroica palmarum NTM 3.0 3
Song Sparrow Melospiza melodia SD 3.0 3
Clay-colored Sparrow Spizella pallida SD 2.0 2
Cape May Warbler Dendroica tigrina NTM 2.0 2
Common Redpoll Carduelis flamnea SD 2.0 2
Blackburnian Warbler Dendroica fusca NTM 1.0 2
Yellow-bellied Flycatcher Empidonax flaviventris NTM 1.0 1
American Goldfinch Carduelis tristis SD 0.5 1
Savannah Sparrow Passerculus sandwichensis SD 0.5 1
† Residency status indicated for Neotropical migrants (NTM), short-distance migrants (SD), and residents (RES) (after
Godfrey 1986, Anonymous 1991).
‡ Weighted abundance over 3-yr study period (1993–1995) (see Methods for explanation).
§ Total number of detections over 3-yr study period (1993–1995).

1997 Schmiegelow Birds Resilient Fragmentation Short-Term

Uploaded by

Copyright:

Available Formats

1997 Schmiegelow Birds Resilient Fragmentation Short-Term

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1997 Schmiegelow Birds Resilient Fragmentation Short-Term

Uploaded by

Copyright:

Available Formats

Ecology, 78(6), 1997, pp.

ARE BOREAL BIRDS RESILIENT TO FOREST FRAGMENTATION? AN

Abstract. We studied the effect of habitat fragmentation on the richness, diversity,

INTRODUCTION patches of older forest are surrounded by regenerating

conservation solution (e.g., Mann and Plummer 1993), STUDY AREA

All species Probable breeders only

FIG. 4. Log series a diversity of forest birds in treatment

fect on seven species in the isolated fragments and six

Controls Isolated fragments Connected fragments

DISCUSSION observed elsewhere. We observed no net loss of species

Isolated fragments Connected fragments

White-throated Sparrow SD 1 0.053 NS NS NS

Red-breasted Nuthatch RES NS 2 (0.004) 2 0.022 NS

Common name Scientific name Status† Abundance‡ Detections§

You might also like