Bird Communities after Blowdown in a Late-Successional Great Lakes Spruce-Fir Forest

2005·7.2 MB·English

by  John M Burris

Wilson Bulletin 117(4):341-352, 2005
BIRD COMMUNITIES AFTER BLOWDOWN IN A
LATE-SUCCESSIONAL GREAT LAKES
SPRUCE-FIR FOREST
JOHN M. BURRISi2 AND ALAN W. HANEYi
—
ABSTRACT. In 2001 and 2002, we inventoried the bird communities and vegetation of two 6.25-ha plots
in a late-successional spruce-fir (Picea mariana-Abies balsamea) forestofnorthern Minnesotathatwas severely
disturbed by a 1999 windstorm. We compared these results with those from two nearby plots that were largely
unaffected by the storm. Using vegetation data collected from one ofthe two plots in each location before the
disturbance in 1996 and 1998, we examined similarities between plots before and after the storm. The most
significanteffect ofthe stormon vegetation was a >80% decrease in tree coverand a >100% increasein shrub-
layer structure because of trees that were tipped over or snapped off. Of 30 territorial bird species, 9 held
territories exclusively in the blowdown, while 2 held territories exclusively in the control. By foraging guild,
10of 11 (91%) species ofground-brushforagers had more territorycoverin theblowdown, while7 of 13 (54%)
species of tree-foliage searchers had more territory cover in the control. Black-and-white Warbler (Mniotilta
varia). Chestnut-sided Warbler (Dendroicapensylvanica). Mourning Warbler(OporornisPhiladelphia), Yellow-
bellied Flycatcher (Empidonaxflaviventris), and Red-eyed Vireo (Vireo olivaceus) had significantly {P < 0.05)
more territory coverin theblowdown, whereas Blackburnian Warbler{Dendroicafusca). Golden-crownedKing-
let (Regulussatrapa), and Yellow-rumpedWarbler{Dendroica coronata) hadmoreterritorycoverinthecontrol.
Canonical correspondence analysis revealed that differences in avian territory cover were primarily attributable
to changes in vegetation structure, in particularthe increase ofstructural debris on the ground and the reduction
in tree canopy, occurring because ofthe wind. Received25 October 2004, accepted 30August2005.
—
Forest composition and structure in the Up- winds in excess of 145 km/hr disturbed ap-
per Great Lakes region is greatly influenced proximately 200,000 ha in northeastern Min-
by disturbances, primarily fire, insect out- nesota (USDA Forest Service 2002). We doc-
breaks, logging, and wind (Van Wagner and umented the effects of severe wind distur-
Methven 1978, Bonan and Shugart 1989, Ber- bance by comparing post-disturbance vegeta-
geron 1991, Drapeau et al. 2000). Although tion and bird communities on two blowdown
the most prevalent natural disturbances in this plots with two nearby control plots that had
region are fire and insects, large-scale wind the same disturbance history and vegetation
events that significantly reduce the canopy are structure before the storm. Because bird spe-
believed to occur with average return intervals cies composition is closely related to habitat
of 1,000 years or more (Frelich and Reich structure (Karr and Roth 1971, Willson 1974,
1996, Larson and Waldron 2000, Frelich Niemi and Hanowski 1984, Pearman 2002),
2002). A number of studies have examined and because the wind reduced tree cover by
the effects of disturbances such as fire and more than 80%, with a corresponding increase
logging on avian communities in the Upper in shrub-layer structure and coarse woody de-
Great Lakes region (Apfelbaum and Haney bris from tipped trees and snapped tree-tops,
1986, Schulte and Niemi 1998, Drapeau et al. we expected a community shift from one
2000); however, despite its known impact on dominated by tree-foliage searchers to one
vegetation structure and composition (Frelich dominated by ground-brush foragers. We ex-
and Reich 1996), few researchers have ex- pected responses similar to those following
amined the effects of wind (Smith and Dall- fire (Apfelbaum and Haney 1981, Morissette
man 1996, Dyer and Baird-Phili—p 1997). et al. 2002) and, in some cases, timber har-
On 4 July 1999, a microburst known as a vesting (Hobson and Schieck 1999, Lohret al.
derecho, and characterized by straight-line 2002).
METHODS
Ste'vCeonlslePgoeinotf,NSatteuvreanlsRFe’soionutr,ceWs1, U5n4i4v8.1.ofUWSiAs.consin- We conducted our study in a 200-ycar-old
^Corresponding author; e-mail: black spruce {Picea nuiriana) and balsam fir
[email protected] {Ahies balsamea) forest that originated from
341
342 THE WILSON BULLETIN • Vol. 117, No. 4, December2005
an 1801 stand-replacing wildfire (M. L. Hein- the number oflive and dead shrub stems with-
selman pers. comm.) in northeastern Minne- in 1 m of the right side of the transect. We
sota’s Superior National Forest (Fig. 1). Two used five 1-m^ circular plots centered at 5, 15,
m
blowdown study plots were located on Seagull 25, 35, and 45 along the transect line to
Lake (48° 07' N, 90° 54' W) and two control estimate percent coverofherbs (height <1 m),
plots, minimally affected by the 4 July 1999 exposed mineral (e.g., rock, bare soil), bryo-
storm, were located near Red Rock Bay (Sa- phytes, coarse litter (diameter >5 cm), and
ganaga Lake), approximately 10 km to the fine litter (diameter <5 cm).
northwest of Seagull Lake. Each 250 X 250- We conducted bird surveys on each of the
m (6.25 ha) study plot, surrounded by a 25-m fourplots once per morning foreach of5 days
buffer zone to reduce the effects of edge, was during May-mid-June 2001 and 2002. Sur-
subdivided with flagging into a grid of 50 X veys were performed using a modification of
50-m cells. Using previously collected data Kendeigh’s flush-plot techniques (Kendeigh
from one of the blowdown plots (1996) and 1944, Apfelbaum and Haney 1986). Each sur-
one of the control plots (1998), we employed vey was conducted by one or two experienced
a BACI design (Before, After, Control, Im- birders who plotted on data sheets all birds
pact; Stewart-Oaten et al. 1986, Irons et al. seen or heard from grid-cell vertices. Surveys,
2000, Stewart-Oaten and Bence 2001) to bet- which were restricted to days without signif-
ter illustrate similarities between plots before icant wind or rain, averaged about 6 person-
the disturbance, and changes occurring be- hr, each designed to plot every territorial male
cause of the windstorm. We did not use a using the area.
BACI design to analyze our bird data, how- After the completion of all five daily sur-
ever, because the annual variation in bird pop- veys, bird locations for each plot were com-
ulations is unpredictable (Blake et al. 1994, piled onto summary sheets. Territories were
Collins 2001) and our pre-disturbance avian delineated from clusters of survey registra-
surveys were conducted in different years. tions and other evidence of established terri-
Post-blowdown vegetation surveys were tories, such as active nests, or adults carrying
conducted in 2001 and again in 2002 along food or fecal sacs. We considered likely tran-
50-m transects running through the center of sients, or individuals with territories too large
10 randomly selected grid cells in each ofthe to determine with our method, as visitors (V)
four study plots (n = 4 plots/year X 10 cells/ unless they were recorded in the same location
plot X 2 years = 80). Using the same meth- on at least 3 of t—he 5 survey days.
odology, we surveyed vegetation in one ofthe Data analyses. To address issues ofspatial
pre-blowdown plots in 1996 and one of the dependence within the vegetation dataset, we
control plots in 1998 (n = 2 plots X 10 cells/ first eliminated repeatedly sampled grid cells
plot = 20). Tree and shrub coverforeach spe- while balancing sample sizes between years
cies were estimated using the line intercept and plots. Of the 100 grid cells for which we
method (Canfield 1941). Trees were defined had vegetation data, we retained 62 cells (10
as stems standing <45 degrees from vertical pre-blowdown [1996], 10 pre-blowdown con-
with a diameter at breast height (dbh) ^5 cm. trol [1998], 12 post-blowdown [2001], 11
Shrubs were identified as all stems >1 m tall post-blowdown control [2001], 9 post-blow-
and <5 cm dbh or as live trees standing >45 down [2002], 10 post-blowdown control
degrees from vertical. Dead trees were con- [2002]) for further analysis. Next, we exam-
sidered coarse litter if standing >45 degrees ined the resulting vegetation data for normal-
from vertical and snags if standing <45 de- ity (Q-Q plot and Shapiro-Wilk tests) and ho-
grees. After the storm, diameters of all stems mogeneity of variance (Levene’s test) and
>5 cm that crossed the 50-m intercept line transformed data according to Box-Cox plots
were recorded and used to estimate the vol- (Box and Cox 1964) as necessary. Finally, we
ume of coarse woody debris per unit area. used a two-way analysis of variance (ANO-
We estimated tree and shrub density by re- VA) for each habitat variable {n = 19) to ex-
cording the number and diameter (rounded to amine differences based both on plot type
the nearest 5 cm) oflive and dead trees rooted (blowdown or control) and time (1996 or
within 1 m of either side of the transect and 1998, 2001, 2002). If the ANOVA yielded a
Burris and Haney • BIRD COMMUNITIES AFTER BLOWDOWN 343
M100 150 200
Kilometers
FIG. 1. Location of the study area and the blowdown area in northeastern Minnesota's Superior National
Forest. The blowdown occurred 4 July 1999, a result of' a >145 km/hr microburst.
" ' ;
344 THE WILSON BULLETIN • Vol. 117, No. 4, December2005
LT O
^ON -2^(OsaU T3^o^o3 ^St^Doi-^oo go.H ^s Is (NrD^NoNOoOOct'N^i^otrsO^OT''ii--^^arrso^r'o^^so‘HonoHoNuoooto^r.i'o«<ocsrtdTo^ro'"ro^T^rNqsooorqo'NoOq*0rfo*(mOrNsf(iNN+fttN^frq*Oo^'r*q"C^oQT'i+*"^tnTi'*P^cQtit0^0ondTOt^rroo0qr0oqOddOs
s: ^03 ci/3 Hs ^O i '++1vtD+O—1^sr+’01'++t1t((+N1NO0+1t0^0+1sNO+O1-O+H1(NN+D1O0+O01(f+1NNf(+01Nt(+1N^(^+N1^'+(^1NND+r1'^v+t1Dr^+01O0r+1s40—+1sii+O1O-+N^1Os+rO1^N+mD1O++\t1+r+t1^Or+1(^NO+1((NN+1
^nj; ’2“ B"8 s II ^II f<-3 ro^^\Oinr4 ro Ttco— (N 'OrocooO'-H(N
^ ^ '3 B ^ Q
c/3 > C§/3 2(U '^-Sg 2S o's-'+i't^-(+NoOstT^i'-^N'Dnrf-rOt^t+^toNoOo>spn'(-N+r‘OnOoso o'^r*0-+-P*t3t0^0o^+0t*^ofrP*Q0-+rt^<pO!psIT^r)^om^^p(tN^Op—+iito
iilBo NO'Ocn(Moot^'^roOOroinm+t'^fO»n'r >nro-^(N'':t'^'-^-+'— (NOO
Is 3
X)<N +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1
Sg|t^s |8£! 8§ °§ NcirsoiirrO\-iNo^''o^-+N+»Ot/tOo^ss£OO)((sXNO)o4Nr^r^^(fNo^irTnotN»-OHonfomSrol^orcoono(cNso'<O^Ns(f'NO^—''^v^O^'oc^Oo(‘-Nn^oosr\o-'mooOirmn'rOfoso>^rcnoNooro ro
s||S
i
3^o^o .SwtSc2;£i-? '"gogC2 ’f*>Sl=j).5"4§;?p^IIQ O“iIIi e J IXS^ pfur+oo10iPm+N1nDr((+ONN1j0co+01ocION+NO1^ssOO(+d1^Nit+mn1^or+jOs1(ii+Nn1nm^O+Oo1NNo0>+On1^(+(Xr<*s+t41Xifq)<+^NN1orNOq+-1O'HOd+--1OHror+f41loo-d+ir1^n'dN+NO10Os++Oit'1(r'd+Ni1^ON(r+j-1ON^o+++t1tot^odN+10sod0+Om1 p+1 p+1 —+1 +1
-mShcoo0i= '^-0aSo<0Dc0n3h3TE8&^33^.’^-^S_o1sc>/^3 2£^^so2Snexgw^•>o>(cU^-nX.'>883 •T-jcOO<^3cOO/v3 -£wc^i>/5>3 OcIuIsofOImmPO^lI('uN-OIqIHNms(u'oNsrqqf^'cOuocoOftI5qN^IruooqOIqq'rI'ufoc(tlIqNcrsauo|N’^q-TOcOO3mO^•u—--OOO'O=;IH^’7q^FOcQ~(T3ONt—uOOOi<^toq^^c0(s.^OoOu—c2sN-'Oqpc^0^co.N^-Ou^Oor-rFopqoo0^^N3r-OOu—^oC’NCFOq-OSo^-p Ou2O? qqcoF< d-u2§iP-i d2-acc d d
O> -«a ^0 o ^O ^^ ^2
Z ."^S ^Vor--O4(U3HII ^ 0d td(oM 'd't >d(nN 0d 0tddo" d 0d--0H 00d (dmN 00d +0dt (diNn 0Od0N 0(d0N
o>^, cct2^+c^ T-3 g Xu Uh WpNq V V
C-4O!(-/u<i JCCS59/^3 > ,0 .cic(^/5-U33H >o^O-rfD0^)tC0j 0qd0 m+dt d cdn ods O(dNs sd(oN 0pd0 OmOsS (pdN Oc(NnN OqdN sqo (d0N 0q
3no ^os g“ «oc cc8>v=s(u oi8>-3^
"
O
TCc~33 ^0o-3 cF0^/33 Vc<oshuu ^^r. Ooc/3 ^c/3
cw^^/1D ”_FoF8C 5^.B<+s2U iS§ p§OI^I ^HnII ^^Q3F >II Qj O(dN O+dt 0-d0p 0dp0 —pdI 0pN0O crd-n dOs
•C I ^ §)•£ U ^
§1 £-" g^JI
g o - ^ Q
^is II 8|
.S^<>cW(-Io2uU)q^8^2oC'c/3D qo.-V5„0FS2E3 ^q<2O(^.>3, 5.'^Cc22§e-20«6 _^^^(aOhI5IU C^-I,FF/>P_FD (2(85U0g2HQq
-Sr'H'lg#
SC^^l<2Qr"«c2"3sc£qci"Sc2 p.p(2u Up^D.uP0WDUiC0i .c2ii T0qqu3 0t0duu pdEuH 0uqdqX Tugdq§3 0(UqXq75 •(UqQXq075 d(UqXq75 *JdqdUHOq *udQqdHq TuddqdH3
-^^
Burris andHaney • BIRD COMMUNITIES AFTER BLOWDOWN 345
significant interaction, indicating that the
P3
Ioocgg d(N dOiNn d*in 00N*°O dON <OiNnn dr0n 0dm dbilfofewrdeontwlny waintdh ctoinmter,olweplcootsnduwcetreedcmhaainngienfg-
C3oL.fN (+T1) (+N1 —+1 q+1 dq+1 Nq+O1 (r+-N1< dq+1 tfewcetesnanpallotysteyspetoienxaamgiivneenbyoetahrdainffderdeinfcfeesrebnec--
in NO -H es between years within each plot type. To
control for Type I error across the two simple
* *CQ main effects, we used a Bonferroni correction
c -§ d in qd in cdnI NO qq procedure (Winer et al. 1991) and set alpha
1 .28 00 for each simple main effect at 0.025. If the
cx0u> d+1 q+1 N+O1 00+001 qid+n1 qii+nn1 d0^+01 0d+01 lsiomwp-luepmpaaiinrweifsfeeccto(mtpiamrei)swoanss sbiegtniwfeiecannt,19f9o6l,-
1998, 2001, and 2002 were performed using
Tc3a q dr- od0*n O*q<N 0N0O qdr- q(ONN (0N0 c(aa0n.Bt0o2nc5f/h3ea)nrgretoo.nii-daedntjiufsytetdimealppehraiodsestoaftsi0g.n0if0i8-
|8
+1 +1 +1 +1 m+1 +1 0+1 +1 Because we wanted to correlate bird pres-
X£ (N — NO (N O0N ence with habitat characteristics, we analyzed
(N '-N
our bird data at the same scale as the vege-
tation data (50 X 50-m grid cell), rather than
a C c C c at the 250 X 250-m plot level. This was ac-
00- Uc00 Tl3§U0 .I2fU§ 1>0 Uo12? "qs0O dciosmtprliibustheeddbbyybostelhecyteianrg(422001gr,id20c0e2l)lsaenqduapllloyt
CQ PQ CQ between the blowdown and control plots. To
mitigate issues of spatial dependence, we re-
0 quired all ofthe selected cells within the same
u 0, d 0d (d(Nn (dN year to be a minimum of 50 m apart, and we
HE V did not select the same cell in successive
X years. So that we could later perform a joint
k. rITi) 00q6 0q q wanealyfsuirsthuersinrgeqbuoirtehdbitrhdatansdelevcetgeedtagtriiodn dcaetlal,s
were those for which we had also collected
m vegetation data in the same year. After cell
0O0N0 00 0(N 0(n selection, we recorded by species (based upon
0. d d d d our territory maps) the percentage ofeach se-
<
o>z< HEi> _ mleacrtyedpcuerlplosceosv,ersepdecibeys awetreerriatsosriy.gneFdortosufomr--
Li, r(iN 0r-0' qd d? O galgeianngerg)uialcdcso(red.ig.n,gttroeet-hfoosleiadgeesscerairbcehderb,ytiBmobcekr
a,
a and Lynch (1970). Next, we tested these data
!!. d rdO*’'i d qdvO X••Oo)o5 afonrd huosmeodgeaneointey-woafy vAarNiOanVceA(tLoevteenste’fsortetshte)
its effect of disturbance. Although somewhat un-
conventional, distinguishing bird use by mea-
suring the percentage of each cell covered by
Continued. . r'I); OdN d-t ro'•jcC oiiC-l?E^l-£vO atfewlteteerwrniatspolroyrtesaqluoilnroewdae,dfgiuinsevretnoscdtaehlteee—cptaatdncihfiaftneterrseisnbcuoetfse btwehe-e
1. 3 =^2C2C'' landscape following the blowdown. We are
TABLE ^cOWSOuus>j £QUD(QUJC-J C—>< >3cXDnC 3>DHC QraaQn£rt#3jiaQ^ft)t; (a(mGHwaiualyxrlleaaefnyfdtehca1tWt9o3tl4ecfr,hria1Wtn9iog7re5eyn,sssSizemietni,tabhlb.uotat1nhu9d8p5avo)Svnhiuaafgnniadnrdtdihena1ngb9si8itl7itat)yt-
346 THE WILSON BULLETIN • Vol. 117, No. 4, December2005
tie difference in average territory sizebetween 4.38, blowdown: 51.6 ± 3.75), was signifi-
plot types (blowdown or control), we con- cantly greater in the control after the wind-
cluded that significant differences in territory storm in both 2001 (control: 42.9 ± 5.47,
cover per cell would likely be the result of blowdown: 23.4 ± 5.07) and 2002 (control:
more territories rather than territories of a 37.6 ± 6.08, blowdown: 7.6 ± 3.23). A sim-
larger size. ilar trend was observed in diameter of live
In examining the relationship between hab- trees (LIVDIA) in the blowdown area: mean
itat structure and bird species composition, we diameter decreased by 2002 (7.9 ± 0.86) to
used only the 42 grid cells (21 blowdown, 21 only halfthat observed before the storm (15.6
control) from 2001 and 2002 for which we ± 1.49). Whereas it was not significantly dif-
had both vegetation and bird data. First, we ferent before the storm, evergreen tree cover
used a Pearson correlation matrix along with (CTREE) and shrub or tree cover (CVR) were
principal components analysis (PCA) to min- also significantly greater in the control than in
imize redundancy within the dataset, follow- the blowdown after the disturbance. On the
ing the recommendations of ter Braak (1986, otherhand, percent shrub cover (CSHRB) was
1994) for subsequent canonical correspon- significantly greater in the control before the
dence analysis (CCA). If ^2 variables were blowdown (control: 44.1 ± 5.06, blowdown:
strongly correlated (r > 0.60) within the cor- 22.6 ± 2.04), but was not significantly differ-
relation matrix, we kept only the habitat var- ent afterwards in either 2001 (control: 45.3 ±
iable most strongly correlated with the first 3.96, blowdown: 46.5 ± 5.15) or 2002 (con-
principal component (i.e., the variable ex- trol: 42.8 ± 5.78, blowdown: 32.2 ± 3.45)
plaining a greater amount of the variation due to tipped trees and broken-topped trees
within the data). Next, using the remaining that were still alive in both years. T—he volume
variables (10 of 19), we performed PCA again of coarse woody debris (DEBRIS) the only
to reduce the complexity of the dataset and variab—le that was not measured before the
summarize the habitat variables within the storm was greater {P < 0.001) in the blow-
blowdown and control areas. Finally, we con- down during both 2001 (control: 57.3 ±
ducted CCA, performed by the PC-ORD sta- 12.35, blowdown: 93.3 ± 15.43) and 2002
tistical package (McCune and Mefford 1999), (control: 33.6 ± 8.78, blowdown: 89.4 ±
on the 10 selected habitat variables and 15 14.05).
common bird species (those with territory Of the 30 bird species with identified ter-
cover in at least 10% of the 42 grid cells) to ritories in either the blowdown or control, 18
investigate more closely the relationship be- had territories in both plot types. Two species
tween habitat characteristics and the distribu- had territories only in the control while nine
tion of bird species. To determine the signifi- species had territories exclusively in the blow-
cance level of this relationship (ter Braak down. Seven territorial and visitor species re-
1987), the CCA included a Monte Carlo test corded in the blowdown were not recorded in
on the firsttwo canonical functions, conducted the control, whereas all species recordedinthe
with ,000 permutations and using time ofday control had territories or were recorded as vis-
1
as the source for randomization. Means are itors in the blowdown.
presented ± SE. Species for which we detected a greater
percentage of territory cover per grid cell in
RESULTS the blowdown included Black-and-white War-
Twenty-six percent (5 of 19) of the habitat bler (scientific names listed in Table 2; con-
variables examined in the blowdown were sig- trol: 2.1 ± 1.49, blowdown: 13.3 ± 4.49, F140
n2i0f0i2cantwlhyedniffceoremnptarafetder wtihtehstporrem-sitnor2m001estoir- (=con5t.r6o0l,:P0,=bl0o.w02d3o)w,n:Che1s2t.n1ut±-si4d.3e5d,Warble”r
mates collected in 1996 (Table 1). In contrast, 7.81, P = 0.008), and Mourning Warbler
there were no significant differences in habitat (control: 0, blowdown: 16.2 ± 5.72, F140 =
variables between years (1998, 2001, 2002) in 8.01, P = 0.007; Table 2). Species with a
the control. Percent tree cover (CTREE), greater percentage of territory cover per cell
which was somewhat higher in the to-be dis- in the control included Blackburnian Warbler
turbed area before the storm (control: 40.1 ± (control: 20.5 ± 6.57, blowdown: 3.3 ± 1.90,
2 »
^ ; H
Burris and Haney • BIRD COMMUNITIES AFTER BLOWDOWN 347
c3
g £ oOr1<0^di—n'dCNMO d^incdOnN-dC^NdcONdO00Old^^CcdNnOdOcNdnOcdtMnCd^M dtn diNnO dCO(TM) d0 dc(nndcMdNOd(NOdOOdN'd^idndONd(NONd^NOod(oN
w
(/)
(sD zC3
O
-o .o2 oc-^oNrid^nd'o^OoNtrD^^O'—O'^Or—N^O^opo^cdmMd (N ^(N ONO (OtN^Mtod^odoqooccdNoOoi-nd^ocTNMfNOpO^iOnNN0c0nO(0cN0nON
T3 3
^
-C
> S
bo <^2 >• m On 00 (N ON ^ <iNn (noo—'(sicntnminONCMO—'cn
^ Uh
e 2
a ^ C^/5 i^0n0 ^—' o NpinOpTiptn'd'O^cdcnNTimj-‘n--Nh^’Drc^ndicMniron^N^cindn 0rd-\H dOTtN OTitOn^CctMn c-—oi^T^dtp—-^itHfCO^M'p^otrdn^oTdotcdONNdOOrdO^
SOo ^ V 00+601 Cd+M1 ^d+1 f—+d;1-C+3M1;fO+d1Od+p1-C+^M1pd+T1td+(1Nd+1CNd+1ONd+1NdO+1 > 0d+\1 cd+n1 0+01 —(+M1 > ^d+1ro+^1pd+1^t—+ico+M1Od+1'^d+1Od+t1nd+1ind+1
-c -2 ' cn
E
3
C 3
—
§0>..^= O3 (OiNn CdiMn Pdin P0d0 Oidn C(iNMn OTdtN d—I CNdMO (iNn mdCM oidn 0d0 d NO
iS^I21'E§ N+O1 o+1 o+1 +1 I I I > +1 I Od+1 pd+1 (—+N1 I d+1 d+1 dd+1(N+MO1-d+H1pd+1r^d+1—0«+01 > oO+N1 Oi+n1
5 On O (N
y-v^ OQ-nn T3
PU C
2
+1 -J
=ao—> h:1/:3 13; >CUU-O- <J[UX, Ju£U g^Scin<l<Z„U^O|aZ.SoU.c£ni^. OCzXCOuCZaOc>^P>Q uuXX oumX Ou uX H^w< S^IIU^ d<> X^< U<
”0
u 3 o
o
o> c00 ^^
cj -r ^o
2U>.,2I0 2V5
_ o "d
3O 3C -2 2 ^
rU^-(f^(f=XgO33f§a2iiJ)NNNl>-01r(><_£—<_O>3s33^jNC3—s•sI-O3o^<,oAU 2—>"X^o<03u^—"<_-oaOa35-_x,Xu3oo^J:> I>iI^2U>'Idc2-^O!3n/5i XH^1^ES222c:Z22^3w§~^^3N,j UX•O^-Ia2£v2P5 5lO"-A->3w^Si22a22lZ-=tt2oE3^5a-r Sri•iC«S222=? l^-I-mo§h^2tSV<i/^.a^£^5?£2I2.5|^"*2E§>ISr: HI23 ia3i P^"•"23u^^^2C522 X-^I^§^cU>2c2otr5n c X>no§-2gX2iJ-2'.‘(2§2O?C2U•Xl^"o2 O--.d2223^cwE^o2oEjX,d—">^2ow65E12t):$2Ea=>.c2^^Is?! ^^2^2cis22_?uZd2‘,3^o2^^5j >2^>2Xrco^CEEa-^ X2i2^XE3o> u^^^2X^^33
348 THE \STLSON BLXLETIN • Vol 117. So. 4. December2005
T.ABLE 3. Selected habitat variables and associated correlations vrith each of three principal components
ha\ing eigenvalues >1. PC.X based on 2001 and 2002 data from 21 blowdo\\Ti and 21 control cells. Superior
National Forest, Minnesota.
\-2n2bie PC : PC: PC3
^ tree cover 0.43 -0.08 -0.08
No. dead trees/ha 0.17 -0.33 -0.54
Live tree diameter tcm» 0.40 -0.10 0.34
^<c shrub cover 0.35 -0.12 -0.06
No. live shrub stems-ha 0.32 0.53 0.00
No. dead shrub stems/ha 0.35 0.01 -0.18
‘T herb cover 0-17 -0.46 -0.06
bryoph>te cover 0.31 -0.14 -0.16
^ coarse liner cover 0.18 -0.31 0.72
Coarse woody debris tm'/hai -0.24 -0.51 -0.02
F = 6.28. P = 0.016). Golden-crowned The Monte Carlo permutations test con-
Kinglet (control: 16.2 = 4.62. blowdown: 0.7 ducted with the CCA indicated that both the
= 744. = 11.03. P = 0.0021 and Yellow- first canonical function (P = 0.027) and the
rumped Warbler (control: 9.0 = 3.41. blow- overall test (P = 0.010) were significant, with
down: 1.0 = 7.42. F — 5.38. P = 0.026: the correlation between selected species and
Table 2). habitat being relatively high (r = 0.84). The
By foraging guild. 6 of the 14 (43^) spe- first axis of the CCA accounted for 9.9^4 of
cies of ground-brush foragers and flycatchers the variation in the bird (iata. and was posi-
held territories in the blowdown but not in the ti\ely correlated with the volume of debris
control: 6 of the 8 (75*^) species holding ter- (DEBRIS, r = 0.51 ) and negatively correlated
ritories in both blowdown and control had a with tree cover (CTREE, r = —0.72). Bird
greater percentage of territory cover in the species preferring hea\y cover at or near the
blowdo\^Ti than in the controls. Four ofthe 13 ground with little to no canopy cover (Mourn-
(31*^) species of tree-fohage searchers had ing Warbler. Chesmut-sided Warbler. Yellow-
more territory cover in the control (all P < bellied Flycatcher, and Winter Wren) —were
0.05). Only the Red-eyed \'ireo had a greater positively correlated with th—e first axis the
percentage ofterritor\ cover in the blowdo\sTi volume ofdebris in particular and are shov^m
(control: 2.1 = 2.14. blowdown: 12.6 =: 4.23. in the extreme right hand portion of Figure 2.
= 4.87. P = 0.033: Table 2). Species such as the Golden-Cro\\Tied Kinglet,
Three principal components had eigenval- Blackburnian Warbler. Swainson's Thrush,
ues >1 (PC 1 = 3.71. PC 2 = 1.78. PC 3 = and Northern Parula were negatively correlat-
1.18) and together explained 67^ of the var- ed with the first axis and preferred more tree
iance in the vegetation (iataset. The first prin- cover (Fig. 2).
cipal component explained 37^ of the vari- Although not significant, the second canon-
ance and was positively correlated with the ical function explained 5.0*4 of the variance
diameter of live trees and tree cover, while in the bird data (Monte Carlo test: P = 0.21)
being negatively correlated with the volume and was positively correlated with bryoph>te
of debris (Table 3). The second component, cover iCBRYO. r = 0.66) and herb cover
which explained 18*^ of the variance, was (CHERB. r = 0.41). Birds most closely as-
positively correlated with the number of live sociated with bryoph\t:e and herb cover in-
shrub stems and negatively correlatedwith the cluded Nashville Warbler. Northern Parula,
volume of debris (Table 3). A plot of PC 1 and WTiite-throated Sparrow.
versus PC 2 (not showm) revealed only slight
DISCUSSION
overlap of blowdown and control cells, indi-
cating that the 10habitat variables retainedfor Our data suggest that the primary effect of
use with the CCA reasonably separate one the 4 July 1999 storm was a sigruficant de-
t>pe from the other. crease in tree canopy and the diameter oflive
Burris and Haney • BIRD COMMUNITIES AFTER BLOWDOWN 349
O)
X
Low 4 Volume ofdebris » High
High 4 Tree cover and live tree diameter Low
F 1
FIG. 2. Bird distribution and vegetation variables (2001, 2002 data) based on functions 1 (FI) and 2 (F2)
ofa canonical correspondence analysis of 10 vegetation variables (codes defined in Table 1) and 15 bird species
(codes defined in Table 2) from 21 blowdown cells and 21 control cells following acatastrophic 1999blowdown
in a black spruce-balsam fir forest in the Superior National Forest, Minnesota. The length and direction ofthe
vector for each habitat variable corresponds to the level of its correlation with each function.
trees, with a concomitant increase in shrub amount of cover at or near the ground in the
layer structure and coarse woody debris. Tree blowdown area. Coarse woody debris in the
cover, which was generally characterized by blowdown area also increased significantly as
black spruce, balsam fir, and paper birch (Bet- a result of the storm.
ida papyrifera), was slightly greater in the Many researchers have documented the im-
pre-blowdown but reduced to half that of the portance of coarse woody debris to avian
control as a result ofthe windstorm. The wind communities (Davis et al. 1999, Greenberg
also decreased the number of live trees and and Lanham 2001, Lohr et al. 2002), citing
the diameter of both live and dead trees by increases in nest-site suitability and food
blowing over or breaking off all but the larg- availability as possible explanations (Lohr et
est dead trees and most of the bigger live al. 2002) for its importance. Chestnut-sided
trees. In the shrub layer, fallen trees and tree- and Mourning warblers, which were strongly
tops eliminated disparities between distur- associated with the volume of coarse woody
bance and control plots with respect to shrub debris, are often associated with dense shrub-
cover and the number of live shrub stems that bery and open woods of early successional
existed before the storm by increasing the forests (Apfelbaum and Haney 1981, Ehrlich
350 THE WILSON BULLETIN • Vol. 117, No. 4, December2005
et al. 1988, Schulte and Niemi 1998). Winter ers and ground-brush foragers and fewer tree-
Wren was also associated with the low-canopy foliage searchers (Apfelbaum and Haney
blowdown despite being typically associated 1986, Drapeau et al. 2000, Morissette et al.
with old-growth forests (Hejl et al. 2002). Yel- 2002). The effect of logging on bird com-
low-bellied Flycatcher, White-throated Spar- munities is largely dependent upon the num-
row, and Black-and-white Warbler also ber ofresidual trees and snags and the amount
showed some preference for areas with higher of coarse woody debris (Brawn et al. 2001,
levels ofcoarse woody debris, with all but the Lohr et al. 2002). Unlike fire or wind, rela-
White-throated Sparrow having significantly tively few snags remain afterclear-cuts, which
more territory cover in the blowdown. Red- leads to a nearly complete change in avian
eyed Vireo, a species often associated with community composition (Schieck and Hobson
closed-canopy or mature forest (James 1976, 2000, Brawn et al. 2001). Natural disturbances
Faanes and Andrew 1983), also had signifi- like wind, and arguably timber harvests in
cantly more territory cover in the blowdown some cases, result in more heterogeneous
but has been shown to respond better than ex- landscapes as a result of different serai stages
pected to canopy loss (Greenberg and Lanham (Niemi et al. 1998), thereby enhancing the di-
2001, Faccio 2003). versity ofbird communities (Angelstam 1998,
Golden-crowned Kinglet and Blackburnian Brawn et al. 2001).
Warbler had significantly more territory cover
ACKNOWLEDGMENTS
in the control than in the blowdown and were
highly correlated with the overall amount of We are indebtedtoE. M. AndersonandT. E Ginnett
tree canopy cover and the diameter of live who suggested improvements to the manuscript. Pro-
trees (Fig. 2). Both species typically forage, ject support was provided by E. Linquist with field
and spend most oftheir time, high in the trees assistance from R. L. Anderson, S. I. Apfelbaum, T.
(Ehrlich et al. 1988, Morse 1994), and their D. Bischof, J. I. Burton, J. Carlson, L. A. Ebbecke, E.
numbers would likely decline if that stratum JM..L.ErnKsrto,ll,A.L.L.B.GrKarholalm,,EJ.. JB..LGairna,haSm.,M.E.LBe.hnHhaanredyt,,
were reduced. J. J. Nagel, S. E. Nielsen, R. Power, and J. D. Thorn-
Overall, a significant decrease in tree can- ton.
opy cover and the volume of coarse woody
debris have provided more opportunities for LITERATURE CITED
species that forage or nest (or both) at or near Angelstam, P. 1998. Maintaining and restoring bio-
the ground, while limiting opportunities for diversityinEuropeanborealforestsbydeveloping
species more likely to use tree canopies. natural disturbance regimes. Journal ofVegetation
While these effects do parallel some of the Science 9:593-602.
responses to fire or timber-harvest disturbanc- Apfelbaum, S. and A. Haney. 1981. Bird populations
es, differences are apparent as well. Both wind before and after wildfire in a Great Lakes pine
aerndnfuirmebleerasd toofa dsencalgisn,e ianndtreeancanionpcyre,agsreeati-n Apfebfloibrreadsutpm.o,pCuoSln.adtIoi.ronas8n3dd:u3rA4i.7n-g3H5as4nu.eccye.ss1i9o8n6.folClhoawnignegsfiirne
ground and shrub-layer cover. After fire, how- in the Northern GreatLakesWilderness. Pages 1-
ever, trees often die slowly over several years, 16 in National wilderness research conference:
and, in the Great Lakes region, they may re- current research (Robert C. Lucas, Comp.). Gen-
main standing for several years before con- eral Technical Report INT-212, USDAForestSer-
tributing to the volume of coarse woody de- vice, Intermountain Research Station, Ogden,
Utah.
bris. In contrast, severe wind resulted in an Bergeron, Y. 1991. The influence ofisland and main-
immediate decrease in tree cover and a cor- land lakeshore landscapes on boreal forest firere-
responding increase in shrub-layer structure gimes. Ecology 72:1980-1992.
and coarse woody debris. Like the effects of Blake, J. G., J. M. Hanowski, G. J. Niemi, and P. T.
wind, logging activities also result in a reduc- Collins. 1994. Annual variation in bird popula-
tion of tree canopy and tree stem density, and tions ofmixed conifer-northern hardwoodforests.
Condor 96:381-399.
anSiinmcirleaarsetoinwchoaatrsweewofoodunyddeabfrtiers.the wind- Bocku,laCt.ioEn.saonfdbJu.rEneLdynancdh.un1b9u70r.neBdrefeodreisntgibnirtdhepoSpi--
storm, post-fire bird communities are typically erra Nevada. Condor 72:182-189.
distinguished by higher densities of flycatch- Bonan, G. B. and H. H. Shugart. 1989. Environmen-