ArticlePDF Available

Natural variability in forests of the Grand Canyon, USA


Abstract and Figures

AimCompare contemporary with pre-fire-disruption forest structures, assessing the influence of factors that caused ecological change and evaluating remote sites as relatively natural areas.LocationGrand Canyon National Park contains the largest never-harvested and long-term ungrazed forest ecosystem in Arizona, providing valuable sites for measuring natural variability. However, anthropogenic disruption of natural fire regimes since Euro-American settlement c. 1880 has led to changes in forest structure.Methods We compared species composition, tree structure, regeneration, and canopy cover on large (135–603 ha) ponderosa pine-dominated study sites: (1) isolated points on the North Rim where some surface fires continued after 1880, (2) a higher-elevation North Rim site where fire has been excluded and (3) a South Rim site, also without recent fire, with a paired Kaibab National Forest site. Forest tree structure prior to fire-regime disruption was reconstructed with dendroecological techniques.ResultsBefore fire exclusion, all sites had relatively low tree density (140–246 trees ha−1) dominated by large trees (basal area 9.1–28.5 m2 ha−1), primarily ponderosa pine or pine/Gambel oak on the South Rim. Currently all sites are relatively dense (389–955 trees ha−1, 14.1–41.3 m2 ha−1) but patterns of species composition and regeneration differed substantially with fire regime and elevation. Regeneration at continued-fire sites was primarily through sprouting species, Gambel oak and New Mexican locust, forming a shrubby midstorey under a relatively open pine canopy. In contrast, all fire-excluded sites were dense with seed-reproducing conifer species.Main conclusionsComparison of change caused by climate fluctuation, tree cutting, fire exclusion, livestock herbivory, and wildlife herbivory, suggests that fire regime alteration appears to have played the greatest role. The remote North Rim sites provide a close analogue to conditions prior to fire regime disruption, a contemporary example of the forest characteristics that might have been extant had recent human-caused disruption of disturbance regimes and heavy resource extraction not occurred. They merit broader study of natural variability on a range of ecological variables in ponderosa pine ecosystems.
Content may be subject to copyright.
Natural variability in forests of Grand Canyon,
Peter Z. Fule
, W. Wallace Covington
, Margaret M. Moore
, Thomas A. Heinlein
and Amy E. M. Waltz
Ecological Restoration Institute and
School of Forestry, Northern
Arizona University, P.O. Box 15018, Flagstaff, AZ 86011, USA
Aim Compare contemporary with pre-®re-disruption forest structures, assessing the
in¯uence of factors that caused ecological change and evaluating remote sites as
relatively natural areas.
Location Grand Canyon National Park contains the largest never-harvested and long-
term ungrazed forest ecosystem in Arizona, providing valuable sites for measuring
natural variability. However, anthropogenic disruption of natural ®re regimes since
Euro-American settlement c. 1880 has led to changes in forest structure.
Methods We compared species composition, tree structure, regeneration, and canopy
cover on large (135±603 ha) ponderosa pine-dominated study sites: (1) isolated points
on the North Rim where some surface ®res continued after 1880, (2) a higher-elevation
North Rim site where ®re has been excluded and (3) a South Rim site, also without
recent ®re, with a paired Kaibab National Forest site. Forest tree structure prior to ®re-
regime disruption was reconstructed with dendroecological techniques.
Results Before ®re exclusion, all sites had relatively low tree density (140±246 trees
) dominated by large trees (basal area 9.1±28.5 m
), primarily ponderosa pine
or pine/Gambel oak on the South Rim. Currently all sites are relatively dense (389±
955 trees ha
, 14.1±41.3 m
) but patterns of species composition and regener-
ation differed substantially with ®re regime and elevation. Regeneration at continued-®re
sites was primarily through sprouting species, Gambel oak and New Mexican locust,
forming a shrubby midstorey under a relatively open pine canopy. In contrast, all ®re-
excluded sites were dense with seed-reproducing conifer species.
Main conclusions Comparison of change caused by climate ¯uctuation, tree cutting,
®re exclusion, livestock herbivory, and wildlife herbivory, suggests that ®re regime
alteration appears to have played the greatest role. The remote North Rim sites provide a
close analogue to conditions prior to ®re regime disruption, a contemporary example of
the forest characteristics that might have been extant had recent human-caused
disruption of disturbance regimes and heavy resource extraction not occurred. They
merit broader study of natural variability on a range of ecological variables in ponderosa
pine ecosystems.
Forest structure, ®re, Arizona, ponderosa pine, mixed-conifer, aspen, reference condi-
B0 700655
Journal No. Manuscript No.
Dispatch: 20.12.01 Journal: JBI
Author Received: No. of pages: 17
Correspondence: Peter Z. Fule
Â, Ecological Restoration Institute, P.O. Box No. 15018, Flagstaff, AZ 86011, USA. E-mail:
Journal of Biogeography, 29, 1±17
Ó2002 Blackwell Science Ltd
Aldo Leopold's (1941) seemingly straightforward concept of
an ecological `base datum' has expanded into a complex
debate over the role of natural variability in developing
conservation strategies. Natural variability [also: `natural
range of variability', `historical range of variability,' or
`reference conditions,' (Landres et al., 1999)] is most com-
monly de®ned in terms of past conditions. For instance, in
western North America, Stephenson (1999, p. 1253) de®ned
reference conditions as `the spectrum of ecosystem condi-
tions ¼within a de®ned area over a speci®ed time period
preceding Euro-American settlement.' Recent human-caused
change is the root cause of the world's major environmental
problems, from extinctions to climate change. Given the
pervasive impacts of modern industrial society on global
ecosystems, there are clear advantages in studying the
relatively unfettered past patterns of ecological structure,
function, and disturbance, as mediated by resource manage-
ment practices of non-industrial societies (e.g. Denevan,
1992; Anderson & Moratto, 1996). Past conditions are
extraordinarily important for understanding the evolution-
ary environment (Millar & Wolfenden, 1999; Moore et al.,
1999) and measuring ecological degradation. But the past is
also problematic. The accuracy of reconstructed conditions
can be uncertain (Stephenson, 1999; Tiedemann et al.,
2000). Climate change, species extirpations or introductions,
and different social priorities may reduce the modern
relevance of historical data (Millar & Wolfenden, 1999).
Finally, stochastic, historically contingent, or non-equilib-
rium events in the past may have created unique and
unrepeatable conditions (Swetnam et al., 1999). Stephen-
son's (1999, pp. 1254±1256) de®nition of `natural' linked
past with contemporary conditions: `the (dynamic) condi-
tions that would exist if the dominant Euro-American
culture had never arrived, but Native Americans had
continued to use the landscape.
South-western ponderosa pine forests (scienti®c names of
species are given in Table 1) provide examples of alternative
approaches to characterize natural variability. Forest struc-
tures and frequent-®re disturbance regimes were disrupted
after Euro-American settlement throughout the south-west,
leading to increasingly large and intense wild®res (Leopold,
1937; Weaver, 1951; Cooper, 1960; Covington & Moore,
1994; Swetnam & Baisan, 1996; Swetnam & Betancourt,
1998). Retrospective analyses of ecological change in pon-
derosa pine forests have included ®re-scar reconstruction of
®re regimes (Swetnam & Baisan, 1996), pre-disruption
forest structure (Covington & Moore, 1994; Fule
Âet al.,
1997), regeneration (White, 1985; Savage et al., 1996; Mast
et al., 1999), and historical evidence from photos, early
twentieth century inventories, or early forest plots (Minnich
et al., 1995; Moore et al., 1999). At least four complement-
ary approaches are being used to increase the resolution
of understanding of natural variability in south-western
ponderosa pine forests:
1 Reconstruct past conditions through dendroecological,
palaeoecological, or historical ecology techniques.
2 Measure relatively undisturbed contemporary sites to
compare with reconstructed data and to explore the
effects of altered modern conditions such as high CO
levels. For example, ®re regime and/or forest character-
istics have been measured at relict sites in Zion National
Park, UT (Madany & West, 1983), northern Mexico
Â& Covington, 1997) and El Malpais National
Monument, NM (Grissino-Mayer & Swetnam, 1997).
3 Draw inferences from ecological relationships observed in
disturbed sites. For example, Ganey et al. (1999) drew
inferences about past ecosystems based on observations
of modern habitat use by Mexican spotted owls.
4 Restore natural ecological conditions as a modern model
for observing ecosystem function (Leopold, 1941;
Covington et al., 1997).
Grand Canyon National Park contains the largest never-
harvested forest in Arizona, with 8600 ha subalpine (spruce-
®r) forest, 1230 ha Douglas-®r forest, 37,179 ha ponderosa
pine/mixed conifer forest, 55 ha aspen forest, and 1787 ha
montane meadows (Warren et al., 1982). The apparently
minimal effects of industrial human society in places such as
national parks, wilderness areas, research natural areas, and
relict sites isolated by geography or ownership, should not
be confused with a blanket assumption of pristine conditions
in these places. At Grand Canyon, ®re regimes have been
disrupted across most of the park (Duhnkrack, 1982; White
& VanKat, 1993; Wolf & Mast, 1998; Fule
Âet al., in review)
Table 1 Tree species found on sampling plots at Grand Canyon study sites
Species Common name Code
Abies lasiocarpa (Hook.) Nutt. Subalpine ®r ABLA
Abies concolor (Gordon & Glendinning) Hoopes. White ®r ABCO
Juniperus osteosperma (Torr.) Little Utah juniper JUOS
Picea engelmannii Parry ex Engelm. Engelmann spruce PIEN
Pinus ponderosa var. scopulorum P. & C. Lawson Ponderosa pine PIPO
Pinus edulis Engelm. Rocky Mountain pinyon PIED
Populus tremuloides Michx. Quaking aspen POTR
Pseudotsuga menziesii (Mirb.) Franco var. glauca (Beissn.) Franco Rocky Mountain Douglas-®r PSME
Quercus gambellii Nutt. Gambel oak QUGA
Robinia neomexicana Gray New Mexican locust RONE
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
2 P. Z. Fule
Âet al.
because of herbivory from livestock grazing (c. 1880±1940)
and irrupting deer populations, as well as ®re suppression
(Mitchell & Freeman, 1993). Anomalous tree regeneration
has been linked with regional climatic events in the twentieth
century (Savage et al., 1996).
Nonetheless, the absence of tree harvesting makes the
park a valuable place to study natural variability using any
of the approaches listed above. For the present study, we
used the ®rst two approaches to ask whether forest struc-
tures have changed following disruption of the frequent ®re
regime at low- and high-elevation sites. We measured forest
density, basal area, canopy cover, age structure, species
composition, and regeneration, applying dendroecological
methods to reconstruct past forest structure.
Then we compared these measurements with evidence
compiled from climate research, forest regeneration studies,
historical data on timber harvest, an extensive body of
knowledge on the Kaibab deer herd (one of the world's best-
studied and most in¯uential examples of wildlife manage-
ment), and cultural sources of information on livestock
grazing practices. This information was interpreted in light
of our forest structural ®ndings to address two additional
questions: to what extent can the individual in¯uences of ®re
disturbance, livestock grazing, deer herbivory, tree cutting,
and climate be assessed? Finally, has natural forest structure
been maintained at remote sites where ®re regimes were not
completely disrupted?
Study area
The study sites formed a geographical gradient (island ®
peninsula ®mainland) on the North Rim, including three
remote sites that have maintained the most nearly undis-
rupted ®re regime documented in Arizona (Table 2). Eleva-
tion increased along the gradient from 2256 m (ponderosa
pine/Gambel oak forest) to 2537 m (mixed conifer forest).
To distinguish geographical from elevational effects, we also
selected a low-elevation (2264 m) pine/oak mainland site on
the South Rim and adjacent Kaibab National Forest. The
study sites totaled 1755 ha (Table 2, Fig. 1), all within
Grand Canyon National Park except a 207-ha Kaibab
National Forest site (southern twenty-three plots in Fig. 1).
Soils at the North Rim sites are predominantly of the Soldier
series, derived from Kaibab limestone (Bennett, 1974). Soils
at the Grandview (South Rim) site are classi®ed as ®ne,
smectitic, mesic, Vertic Paleustalfs and Haplustalfs, clay soils
weathered from calcareous sandstone parent material
(B. Lindsay, National Resource Conservation Service,
personal communication, 2000). Average annual precipita-
tion at the North Rim ranger station (elevation 2542 m) is
58.4 cm, with an average annual snowfall of 328 cm.
Temperatures range from an average July maximum of
26 °C to an average January minimum of )2°C. At the
South Rim (elevation 2125 m), average annual precipitation
is 36.8 cm with an average annual snowfall of 177.5 cm;
average July maximum temperature is 29 °C and average
Table 2 Study site characteristics, listed from lowest to highest elevation. The Kaibab Forest and Grandview sites are located on the South Rim. The following four sites form a
geographical and elevational transect from west to east on the North Rim. Fire regime data from Fule
Âet al. (in review)
Study site Code
slope (%) Vegetation type WMPI
* Fire regime disruption
Kaibab KF 207 23 2244±2284 South Rim,
Forest mainland 11.5 Ponderosa pine/Gambel oak 8.94 1887
Grandview GV 603 67 2250±2274 South Rim, 1887, then prescribed burns in
mainland 11.5 Ponderosa pine/Gambel oak 8.94 some portions in 1980s
Powell PP 315 36 2256±2336 North Rim, 1879, then ®res in 1892, 1924
Plateau island 11.5 Ponderosa pine/Gambel oak 8.56 and 1987
Fire Point FP 135 15 2308±2368 North Rim, 1879, then ®res over western 1/3
peninsula 9.9 Ponderosa pine/Gambel oak 6.25 of site in 1923 and 1989
Rainbow RP 225 25 2305±2335 North Rim, 1879, then ®res in 1900, 1985,
Plateau peninsula 23.2 Ponderosa pine/Gambel oak 7.53 and 1993
Swamp SR 270 30 2427±2537 North Rim, Mixed conifer: ponderosa pine,
Ridge mainland 13.3 white ®r, Douglas-®r 8.70 1879
Weibull Median Probability Interval, a probabilistic measure of central tendency in a Weibull distribution ®t to ®re interval data (Swetnam & Baisan, 1996) for ®res
scarring 25% or more of the sample trees. These ®res are presumably relatively large, crossing much or all of each study site. At each site, the WMPI is calculated from c. 1700 to the ®re
regime disruption date listed in the last column.
Disruption date of the frequent ®re regime which prevailed prior to European settlement.
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
Natural variability in forests
January minimum temperature is )1°C (Bennett, 1974;
Sellers & Hill, 1974; GCNP, 1992; White & Vankat, 1993).
All sites were used by native Americans for resource use; all
but the SR site were occupied at least seasonally for farming
(Altschul & Fairley, 1989
2). Resource use by European settlers
± livestock grazing, tree cutting, mining and tourism ± began
around 1870 on the North Rim, 1887 on the South Rim
(Verkamp, 1940; Altschul & Fairley, 1989). All the study sites
had frequent ®re regimes prior to European settlement, but
®res ceased at the `mainland' sites SR and GV/KF after 1879
and 1887, respectively. Fire regimes were also altered after
1879 at the isolated point/plateau sites but each of these sites
had two or three large surface ®res since 1879 (Table 2).
Field methods
Permanent plots based on the National Park Service's Fire
Monitoring protocol (NPS, 1992; Reeberg, 1995) were used
to measure current conditions of vegetation and fuels, and to
collect dendroecological data for reconstruction of past
forest structure. The forest sampling type was de®ned as at
least 10% ponderosa pine forest cover prior to European
settlement, based on the presence of pre-1880 era trees,
snags, stumps, or logs. In mixed species stands, ponderosa
pine must have been a dominant tree in the stand with old
individuals (or remnants) present. Sampling plot origins
were located from a systematic 300 m grid placed over each
sampling site. When a gridpoint fell in an unsuitable location
(e.g. archaeological site), the points 50 m N, E, W, and S
were checked for suitability. If none were acceptable, the
gridpoint was discarded.
Sampling plots were 0.1 ha (20 ´50 m) in size, orientated
with the 50 m sides uphill±downhill to maximize sampling
of variability along the elevational gradient and to permit
correction of the plot area for slope. Plot corners and centres
were marked with 30-cm iron stakes sunk ¯ush to the forest
Figure 1 Study sites on the North and South
Rims of Grand Canyon National Park and
adjacent Kaibab National Forest on the South
Rim. A biogeographical and elevational gra-
dient is formed from the western point/
plateau sites ± Powell Plateau, Fire Point, and
Rainbow Plateau ± to the `mainland' Swamp
Ridge site. The South Rim `mainland' sites ±
Grandview and Kaibab Forest ± are similar in
elevation to the western point/plateau sites
but differ in the recent ®re disturbance
regime. Dark squares are locations of samp-
ling plots.
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
4 P. Z. Fule
Âet al.
¯oor. The distance and bearing from a tagged reference tree
to the centre was recorded. Trees greater than 15 cm
diameter at breast height (dbh) were measured on the entire
plot (1000 m
) and trees between 2.5 and 15 cm d.b.h. were
measured on one quarter-plot (250 m
); all trees were
tagged. Tree attributes measured were: dbh, crown position
(dominant, codominant, intermediate or subcanopy), dam-
age, and tree condition (1. live; 2. declining; 3. recent snag;
4. loose bark snag; 5. clean snag; 6. snag broken above
breast height; 7. snag broken below breast height; 8. downed
dead tree; and 9. cut stump). The condition categories were
derived from snag decomposition studies in dry ponderosa
pine/mixed conifer forests in eastside Oregon and Washing-
ton (Maser et al., 1979; Thomas et al., 1979) and tested in
Arizona (sensitivity analysis described below and in Fule
et al., 1997). Previous research in northern Arizona showed
that the pre-1880 status of trees could be conservatively
identi®ed in the ®eld: ponderosa pines with dbh >37.5 cm
or ponderosa of any size with yellowed bark (White, 1985;
Mast et al., 1999), as well as all oaks, junipers, and pinyon
trees >17 cm (Barger & Ffolliott, 1972). `Conservative'
estimation meant that these criteria included all pre-1880
trees as well as numerous post-1880 trees. Tree status was
later corrected in the laboratory using age data. One
hundred percent of living trees meeting the ®eld criteria
above were considered potentially pre-1880 trees and were
cored. Ten percent of all post-1880 live trees were also
cored. Coring height was 40 cm above ground level, chosen
to meet two objectives: ®rst, to measure tree age, and second,
to measure growth between the ®re regime disruption date
and the present (needed for the forest reconstruction). The
two objectives con¯ict because the best coring height for age
is ground level, but the butt swell and irregular growth
around the root collar make this an inappropriate height for
growth measurement. The 40-cm height is the lowest
position on the bole for consistency of tree form, permitting
a good measurement of growth. Seedling trees, those below
2.5 cm dbh, were tallied by species, condition, and height
class in a 50-m
Canopy cover measured by vertical projection (Ganey &
Block, 1994) was recorded at 30 cm intervals on two 50-m
point intercept transects along the outer plot edges (total of
332 intercept points). Photopoints were established at the
corners and quarter-corners of each plot.
Laboratory methods
Plot areas were corrected for slope. Tree increment cores
were surfaced and visually crossdated (Stokes & Smiley,
1968) with tree-ring chronologies we developed. Rings were
counted on cores that could not be crossdated, especially
younger trees. Additional years to the centre were estimated
with a pith locator (concentric circles matched to the
curvature and density of the inner rings) for cores that
missed the pith (Applequist, 1958). Past forest structure was
reconstructed at the time of disruption of the frequent ®re
regime 1887 at the South Rim sites and 1879 at the North
Rim sites, following dendroecological methods described in
detail by Fule
Âet al. (1997). Brie¯y, size at the time of ®re
exclusion was reconstructed for all living trees by subtract-
ing the radial growth measured on increment cores since ®re
exclusion. For dead trees, the date of death was estimated
based on tree condition class using diameter-dependent snag
decomposition rates (Thomas et al., 1979) or historical
harvesting records for stumps. To estimate growth between
the ®re exclusion date and death date, we developed local
species-speci®c relationships between tree diameter and
basal area increment (r
0.45±0.90). An analogous process
of growth estimation was used to estimate the past diameter
of the small proportion of living pre-1880 trees for which an
intact increment core could not be extracted because of rot.
Contemporary forest structure
Ponderosa pine was the predominant species at each site
(Table 3), making up 73±97% of basal area everywhere
except the high-elevation SR site (48%). Gambel oak was
second in basal area at all the sites but SR, where white ®r,
quaking aspen, and Douglas-®r followed ponderosa pine in
basal area. Other species, especially New Mexican locust
and Rocky Mountain pinyon, were found at several sites but
never represented more than 0.4 m
of basal area. Utah
juniper was found only at the South Rim sites, while quaking
aspen, Douglas-®r, and New Mexican locust were encoun-
tered only on the North Rim. Patterns of tree density
(Table 3) often contrasted with basal area: pines were in the
minority everywhere except the South Rim GV and KF sites.
Oaks accounted for 20±70% of the total trees ha
the sites, except at SR where non-pine species made up 83%
of tree density. New Mexican locust constituted 30% of
trees at the FP site.
Forest age distributions at each site were uneven-aged and
generally dominated by relatively young trees that estab-
lished after ®re regime disruption (Fig. 2). Of a total of 2597
cored trees, 2092 were dated (81%). As all but the KF sites
were never harvested, the age distributions represent unusu-
ally complete records of old trees. Ponderosa pine trees made
up the majority of pre-disruption trees at each site. The
oldest trees by species had centre dates of: ponderosa pine
1537, Gambel oak 1650, Utah juniper 1770, white ®r 1793,
Douglas-®r 1796, aspen 1813, New Mexican locust 1904,
and Engelmann spruce 1932. Sprouting species (Gambel
oak, quaking aspen) predominated in the early regeneration
after ®re regime disruption, generally 1880±1920. Excep-
tions were FP, where very little oak was found, and SR,
where pre-1880 white ®r establishment equalled or exceeded
aspen establishment in nearly every 20-year period. The only
areas where ponderosa pine establishment was dominant
after settlement were the adjacent GV and KF sites,
beginning in the 1921±1940 period.
Regeneration (Table 4) was categorized as trees <30 cm
in height (new seedlings or sprouts), trees between 30 cm
and 2 m in height (established seedlings or sprouts), and
trees >2 m in height, up to 2.5 cm dbh (saplings). The
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
Natural variability in forests
species that reproduce primarily by sprouting, Gambel oak,
New Mexican locust, and quaking aspen, dominated regen-
eration density at all sites except SR. Conifers had substan-
tially lower densities, except for new white ®r seedlings at
SR. Regeneration was highly variable: minimum per-plot
regeneration density was zero at all sites and standard errors
were high relative to means.
Canopy cover ranged from 36.4% at KF (the only site
where trees had been harvested) to 63.2% at SR (Table 5).
The other four sites were within 5% of each other in average
canopy cover. Although the means were similar, per-plot
extremes ranged widely. The RP site had the greatest range,
from the lowest minimum of any site (0.3%) to the highest
maximum (85.5%). Overall variability was greatest at GV,
where the standard error as a proportion of the mean
(17.9%) was over four times greater than that of the next
closest site, RP (4.3%). Canopy cover was correlated with
basal area (r0.65) with a predictive relationship explain-
ing 42% (adjusted r
) of the variation.
Changes since ®re-regime disruption
Although a consistent methodology was used at all sites to
estimate past forest structure, differences in environmental
conditions, species characteristics, and management his-
tory can in¯uence the accuracy of reconstruction and
interpretation of results. Limitations and interpretations
are discussed in `Assessment of reconstruction data',
Increases in total tree density since ®re regime disruption
(1887 South Rim, 1879 North Rim) were statistically
signi®cant at every site except FP (paired t-test, P<0.05);
density increases ranged from 155 (FP) to 581% (GV)
(Table 6). Even without including New Mexican locust, a
small-stature and short-lived species that is poorly suited for
dendroecological reconstruction techniques, the range of
increase in tree density was high: 78% (FP) to 582% (GV).
Presently dead trees (i.e. trees dead in the current inventory
but reconstructed as living at the time of ®re regime
disruption) comprised the following proportion of recon-
structed density: 43 (KF), 32 (GV), 27 (PP), 37 (FP), 29 (RP),
and 58% (SR). Total basal area increases were statistically
signi®cant at every site except KF; basal area increases
ranged from 12 (KF) to 152% (GV). Presently dead trees
comprised the following proportion of reconstructed basal
area: 86 (KF), 43 (GV), 31 (PP), 40 (FP), 37 (RP), and 69%
(SR). The reconstructions were relatively insensitive to
changes in decomposition rates. Comparing the twenty®fth
and seventy®fth decomposition percentiles with the ®ftieth
percentile, reconstructed tree densities varied by 1.9%
(range 0±8.8%) and basal areas varied by 4.8% (range
Ponderosa pine dominated all the study sites at the time of
disruption. Ponderosa pine density was higher in the
Table 3 Forest structure (trees ³2.5 cm dbh) at Grand Canyon study sites in 1997 or 1998. Species codes are derived from the genus and
species (e.g. ABCO Abies concolor). Statistics presented are the mean (standard error), and minimum±maximum
Density (trees ha
KF 689.3 (242.8) 2.6 (1.9) 540.7 (233.9) 146.0 (53.2)
10.0±5481.1 0±40.1 0±5301.1 0±1241.0
GV 955.1 (147.4) 1.2 (1.2) 9.9 (2.4) 3.9 (1.7) 645.7 (140.7) 292.5 (50.2)
120.1±9354.2 0±81.2 0±90.0 0±80.7 20.0±9214.1 0±1893.8
PP 638.1 (102.5) 249.3 (32.6) 289.4 (89.3) 99.4 (45.4)
70.1±2944.9 60.1±953.5 0±2811.5 0±1524.9
FP 388.8 (119.3) 0.7 (0.7) 192.8 (24.0) 79.1 (63.6) 116.3 (86.3)
50.0±1590.1 0±10.3 50.0±340.6 0±946.1 0±1272.0
RP 935.8 (180.0) 6.4 (6.4) 0.4 (0.4) 208.8 (35.6) 7.3 (7.3) 653.2 (175.1) 59.5 (26.1)
100.3±2707.1 0±160.5 0±10.1 10.4±583.5 0±183.4 0±2661.7 0±567.0
SR 940.7 (104.3) 466.5 (62.3) 156.6 (22.9) 255.8 (59.8) 56.3 (19.9) 5.6 (5.6)
309.2±2417.7 0±1132.5 20.2±524.4 0±1411.1 0±475.3 0±168.0
Basal area (m
KF 14.1 (2.1) 0.03 (0.03) 0.02 (0.02) 10.3 (1.9) 3.8 (0.8)
2.1±34.9 0±0.8 0±0.5 0±33.1 0±13.6
GV 22.9 (1.0) 0.002 (0.002) 0.5 (0.2) 0.02 (0.01) 16.8 (1.0) 5.5 (0.7)
6.4±48.4 0±0.1 0±7.0 0±0.7 1.0±44.1 0±26.5
PP 26.3 (1.7) 24.4 (1.5) 1.7 (0.6) 0.3 (0.12)
8.4±51.9 8.4±50.0 0±19.0 0±3.7
FP 31.4 (3.2) 0.01 (0.01) 30.5 (3.2) 0.7 (0.5) 0.2 (0.1)
11.2±56.7 0±0.2 11.2±56.7 0±7.0 0±1.3
RP 27.2 (2.5) 0.06 (0.06) 0.01 (0.01) 21.8 (2.7) 0.007 (0.007) 5.0 (1.2) 0.4 (0.2)
2.7±53.1 0±1.5 0±0.3 0.8±52.8 0±0.2 0±21.4 0±5.2
SR 41.3 (1.9) 14.5 (1.2) 19.9 (1.9) 5.3 (1.0) 1.6 (0.6) 0.004 (0.004)
18.5±62.2 0±26.8 1.6±39.8 0±19.7 0±10.0 0±0.1
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
6 P. Z. Fule
Âet al.
contemporary forest at all sites, but the increase was
signi®cant only at sites GV and PP (paired t-test,
P<0.05). Ponderosa basal area increased signi®cantly at
all but the KF, RP and SR sites. At KF, the contemporary
pine basal area might have been expected to be similar to
that of the adjacent GV site except for the harvesting of
12.5 m
of large ponderosa trees.
Tree density changes as disruption differed by mode of
regeneration, seeding vs. sprouting (Fig. 3). Since disruption,
seed-reproducing pine populations increased substantially
(743±993%) at the two South Rim sites and mesic conifer
species, especially white ®r, increased nearly 1500% at SR.
On the western North Rim, however, where post-settlement
®re regimes were the least disrupted (see Table 2 for the
dates of post-1879 ®res), the densities of pines increased only
slightly to moderately: 65% at PP, 27% at FP, and 34% at
RP. Sprouting species (Gambel oak at all sites but SR,
quaking aspen at SR) increased substantially in basal area
and density at all sites, but the greatest proportional increase
occurred on the western North Rim where oaks rose from 1
to 6% of the pre-disruption tree density to 20±70% of
contemporary density. The proportion of oak basal area
increased much less, as most of the oaks were small in
Stumps provided evidence of past tree cutting at the KF
site, averaging 141.2 trees ha
(SEM 34.8, range
Figure 2 Tree age distributions (centre date
at increment coring height of 40 cm). Black
bars are coniferous species, all obligate seed-
ers. White bars are deciduous species, all
strong sprouters.
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
Natural variability in forests
Table 4 Regeneration density (trees ha
) at Grand Canyon study sites in 1997 or 1998. Statistics presented are the mean (standard error), and minimum±maximum
Regeneration 0±30 cm in height
KF 506.0 (202.1) 17.4 (12.0) 488.5 (200.5)
0±3630.3 0±200.1 0±3630.3
GV 3084.4 (496.1) 6.0 (4.2) 3.0 (3.0) 122.9 (35.5) 2952.6 (495.0)
0±23204.6 0±202.9 0±200.4 0±1411.8 0±23204.6
PP 2596.3 (822.0) 44.9 (21.5) 2238.0 (803.0) 313.4 (106.6)
0±25061.2 0±600.8 0±25061.2 0±3231.2
FP 1757.6 (1035.9) 228.2 (146.1) 905.0 (905.0) 624.4 (312.5)
0±15014.6 0±2204.0 0±13,574.8 0±3509.1
RP 3094.5 (795.1) 158.0 (98.0) 1980.2 (669.3) 956.3 (304.1)
0±13572.4 0±2345.8 0±12361.6 0±5220.2
SR 5304.4 (1032.7) 2397.0 (691.1) 54.1 (25.6) 2185.5 (628.9) 27.0 (21.1) 640.9 (346.8)
0±26833.5 0±13126.8 0±602.4 0±16020.0 0±606.7 0±9011.2
Regeneration 30 cm to 2 m in height
KF 985.9 (457.2) 43.7 (35.7) 942.1 (458.1)
0±7806.2 0±805.7 0±7806.2
GV 700.6 (255.5) 3.0 (3.0) 23.9 (13.8) 9.0 (6.7) 110.7 (57.9) 553.9 (252.2)
0±15203.0 0±202.9 0±800.0 0±400.1 0±3601.6 0±15203.0
PP 2661.4 (547.1) 61.4 (30.8) 1425.5 (480.0) 1174.5 (306.4)
0±13574.8 0±1005.0 0±13574.8 0±3821.7
FP 1659.8 (678.5) 721.0 (380.5) 575.9 (575.9) 362.8 (179.1)
0±8844.2 0±5409.7 0±8638.5
RP 3749.3 (1212.7) 65.2 (43.2) 9.2 (9.2) 1457.5 (845.9) 2217.3 (699.8)
0±27390.5 0±1001.8 0±229.2 0±20542.9 0±13242.2
SR 3411.3 (796.3) 1653.2 (441.6) 40.8 (18.0) 1362.9 (503.4) 6.7 (6.7) 347.7 (197.8)
0±19563.2 0±9479.1 0±406.4 0±10084.1 0±202.2 0±4619.8
Regeneration greater than 2 m in height
KF 69.6 (61.1) 8.8 (8.8) 60.9 (60.9)
0±1400.3 0±201.4 0±1400.3
GV 66.2 (27.1) 6.0 (6.0) 32.9 (21.5) 27.3 (16.5)
0±1400.1 0±400.1 0±1400.1 0±1000.2
PP 559.5 (232.1) 386.5 (201.5) 173.0 (79.2)
0±6010.8 0±5080.4 0±2407.7
FP 333.0 (150.9) 174.9 (91.5) 41.1 (41.1) 117.0 (117.0)
0±1973.9 0±1202.2 0±617.0 0±1754.5
RP 151.9 (84.8) 9.2 (9.2) 41.8 (41.8) 101.0 (49.0)
0±1879.3 0±229.2 0±1044.0 0±835.2
SR 148.3 (70.4) 60.6 (36.4) 74.2 (39.3) 13.4 (9.3)
0±2016.8 0±1008.4 0±1008.4 0±202.2
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
8 P. Z. Fule
Âet al.
0±540 trees ha
). Eighty-nine percent of the cut trees were
ponderosa pine; the remainder was Gambel oak. Trees less
than 40 cm dbh, probably cut in thinning treatments or for
®rewood collection, made up 81.9 pines ha
and 14.5 oak-
. Trees over 40 cm dbh, predominantly pines cut for
lumber production, averaged 43.5 pines ha
and 1.7 oaks
. The total basal area removed in all cutting was
12.6 m
, 98% of which was ponderosa pine
>40 cm dbh. Within Grand Canyon National Park, tree
cutting at the Grandview site averaged 6.5 trees ha
(SEM 1.7, range 0±60 trees ha
. On the North Rim,
the FP site averaged 2.0 cut trees ha
(SEM 1.5, range 0±
20 trees ha
) and the SR site averaged 0.7 cut trees ha
(SEM 0.7, range 0±20.1 trees ha
). No cut trees were
encountered on plots at the PP and RP sites, but infrequently
we observed stumps off plots.
Changes in Grand Canyon forests: assessment
of reconstruction data
We reconstructed past forest conditions on the same sites
where we measured contemporary conditions. Age and past
growth were measured directly on increment cores from
living trees of pre-1880 origin. However, many trees alive at
the time of ®re regime disruption have died. Dendroecolog-
ical reconstruction of past south-western forest conditions is
based on the persistence of dead woody material such as
snags, logs, and stumps from 1879 to present (Covington &
Moore, 1994; Fule
Âet al., 1997). These methods have
recently been tested quantitatively in northern Arizona.
Mast et al. (1999) demonstrated that ponderosa pine wood
Table 5 Canopy cover (measured by vertical projection) at Grand Canyon study sites in 1997 or 1998
Study site N (no. of plots) Minimum (%) Maximum (%) Mean (%) SEM (%)
KF 23 6.7 66.6 36.4 3.2
GV 67 11.4 84.6 47.1 17.9
PP 36 15.4 79.2 49.7 2.1
FP 15 8.7 82.8 51.7 4.2
RP 25 0.3 85.5 48.3 4.3
SR 30 31.6 84.9 63.2 2.2
Table 6 Reconstructed forest structure (trees ³2.5 cm dbh) at Grand Canyon study sites at the date of the ®nal ®re of the pre-European
settlement frequent ®re regime 1887 at the South Rim sites (KF and GV) or 1879 at the North Rim sites. Species codes are derived from the
genus and species (e.g. ABCO Abies concolor). Statistics presented are the mean (standard error), and minimum-maximum
Density (trees ha
KF 144.9 (31.1) 0.4 (0.4) 72.3 (15.8) 72.2 (28.3)
20.0±560.3 0±10.0 20.0±390.1 0±489.7
GV 140.2 (14.0) 8.2 (3.2) 0.1 (0.1) 65.0 (6.5) 66.9 (13.3)
10.1±610.1 0±170.1 0±10.0 0±287.3 0±600.1
PP 157.1 (23.3) 0.3 (0.3) 151.5 (23.5) 5.4 (2.8)
20.2±646.2 0±10.3 20.2±646.2 0±92.3
FP 152.6 (20.3) 151.1 (20.4) 1.5 (1.5)
40.0±310.2 40.0±310.2 0±21.9
RP 159.7 (25.6) 0.9 (0.9) 155.7 (26.1) 3.1 (2.2)
20.0±562.2 0±22.9 20.0±562.2 0±45.8
SR 245.7 (12.9) 31.8 (5.2) 131.5 (10.7) 67.9 (14.1) 14.6 (4.5)
90.1±373.6 0±90.1 20.2±261.6 0±250.3 0±80.9
Basal area (m
KF 12.6 (1.5) 0.01 (0.01) 11.5 (1.4) 1.1 (0.3)
3.3±30.4 0±0.2 3.1±30.4 0±5.4
GV 9.1 (0.6) 0.3 (0.1) 0.0004 (0.0003) 7.9 (0.6) 0.8 (0.2)
0.3±22.8 0±4.7 0±0.02 0±22.1 0±8.2
PP 17.9 (2.5) 0.02 (0.02) 17.8 (2.5) 0.1 (0.05)
4.7±77.3 0±0.7 4.6±77.3 0±1.5
FP 20.5 (2.1) 20.5 (2.1) 0.01 (0.01)
6.5±30.2 6.5±30.2 0±0.2
RP 17.0 (2.9) 0.2 (0.2) 16.8 (2.9) 0.02 (0.01)
4.4±64.5 0±4.7 4.4±64.5 0±0.2
SR 28.5 (1.8) 3.4 (0.6) 21.3 (2.0) 1.1 (0.3) 2.7 (0.8)
15.1±54.0 0±12.0 5.5±49.7 0±6.1 0±15.2
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
Natural variability in forests
persisted since at least 1876 at the Gus Pearson Natural Area
(about 75 km south of the GV study site) and could be
reliably identi®ed in the ®eld as to pre-1880 vs. post-
settlement origin. On adjacent sites, Huffman et al. (2001
reconstructed ponderosa pine densities on 90-year-old-
historically measured plots (trees ³10.2 cm in 1909), ®nd-
ing that errors caused by missed evidence of dead trees
averaged only 3.2 trees ha
, less than 6% of local historic
tree density.
Applying this method in a nearby mixed-species forest,
however, Fule
Âet al. (1997) suggested that it had limited
applicability for New Mexican locust, a small in diameter,
short-lived species. The absence of pre-1880 locust evidence
was interpreted as more likely to have been caused by decay
than to increased locust density in contemporary forests
Âet al., 1997) ± although there is evidence from
elsewhere in northern Arizona that dense locust sprouting
is associated with overstory tree harvesting and ®re exclusion
(Gottfried, 1980). Other species have intermediate charac-
teristics: trees such as Gambel oak, quaking aspen, white ®r,
and the other species encountered at Grand Canyon are
capable of reaching large size and multicentury lifespans. But
these species lack the decay-resistant chemical structure of
ponderosa pine and may have more rapid decay rates,
especially white ®r (Laacke, 1990) and Gambel oak, which
are susceptible to heart rots that often initiate the wood
decay process even before tree death.
Site and disturbance factors also in¯uence reconstruction
accuracy. Decay rates are likely to be less limited by
moisture at the high-elevation SR site. Post-settlement
surface ®res at PP, FP and RP, and prescribed burning of
part of the GV site may have consumed evidence of some
trees of pre-1880 origin. The effect of missing evidence
would be to widen the gap between pre-disruption and
current forest densities. As these sites actually had the
greatest agreement between 1879 and current forest struc-
tures for seed-reproducing species, the conclusions of relat-
ively little change at these sites are strengthened. Utilization
of wood resources and treatment of logging residues have
been limited or non-existent at the park sites but could have
affected pre-1880 evidence at the KF site. Contemporary and
pre-disruption basal area values within sites were signi®-
cantly correlated (P<0.05) at the unharvested park sites
GV, PP, FP and RP (r0.60±0.68), but correlations were
much lower at the SR site (0.47) and non-signi®cant at the
harvested KF site (r0.24). In sum, reconstructions should
be considered most accurate for sites with the least distur-
bance and for species with the greatest resistance to decay
(ponderosa pine, Utah juniper, Rocky Mountain pinyon and
Multiple lines of evidence
Historical evidence from a variety of sources supports the
trend of changes found at the study sites. On the South Rim,
Woolsey (1911) reported a mean of twenty-six pines over
15.2 cm dbh per hectare on `average' stands in timber-sale
areas on the Tusayan National Forest (including the present-
day KF site but also forests to the south around Williams,
AZ) and a mean density of eighty-®ve pines over
10.2 cm dbh ha
on `maximum' stands in the adjacent
Coconino National Forest. Reconstructed pine densities at
the GV site were 59.4 pines > 15.2 cm dbh ha
and 61.7
pines > 10.2 cm dbh ha
; the KF site had 67.1 pines
> 15.2 cm dbh ha
and 71.4 pines > 10.2 cm dbh ha
A 1909 photograph `south of Grand View' shows an open
and apparently unharvested forest structure of large trees
(Fig. 4).
On the Kaibab Plateau, the Lang & Stewart (1910) survey
reported stand averages over a 202-ha area in the ponderosa
pine type of 128.1 pines > 15.2 cm dbh ha
. For compar-
ison, the PP, FP, and RP sites averaged 141.4, 136.4 and
124.8 pines > 15.2 cm dbh ha
in the 1879 reconstruction
and 141.2, 148.5, and 147.8 pines > 15.2 cm dbh ha
the contemporary forest ± nearly no change in pine density
over c. 120 years. An example of the PP forest structure is
shown in Fig. 5. In mixed conifer, Lang & Stewart (1910)
reported density averages for trees > 15.2 cm dbh of: pon-
derosa pine 45.7 trees ha
, Douglas-®r 16.8 trees ha
`balsam' [white] ®r 30.5 trees ha
, and spruce 115 trees
. At the SR site, trees > 15.2 cm dbh in the 1879
reconstruction averaged: ponderosa pine 122.1 trees ha
Douglas-®r 14.2 trees ha
, white ®r 28.1 trees ha
, and no
spruce. The higher pine density and absence of spruce at SR
suggests that Lang & Stewart's (1910) sampling site was at
higher elevation than SR. Other structural studies on the
Kaibab Plateau found 99±111 pines ha
(Rasmussen, 1941)
Figure 3 Relative dominance by seed-reproducing species and
sprouting species in reconstructed 1887 (South Rim) or 1879 (North
Rim) forest and in current forest conditions. Fire-excluded sites KF,
GV, and SR show large increases in seed-reproducing conifer
species. Seeders remained relatively constant while sprouters in-
creased at PP, FP, and RP, where either two or three large spreading
surface ®res occurred after 1879.
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
10 P. Z. Fule
Âet al.
and an average of 138 pines ha
in a dendroecological
reconstruction of 1880 conditions (Covington & Moore,
1992). Moore et al. (1999) noted that pre-1880 ponderosa
pine densities on the coarse-textured limestone soils of the
Kaibab Plateau were about twice as high as on ®ne-textured
basalt soils elsewhere in northern Arizona.
Lang & Stewart (1910, p. 8) noted that `the pine occurs
mostly in open stand [sic] park-like or even isolated in
character¼all age classes in varying density and propor-
tions, but nowhere fully stocked.' They emphasized that
`forest ®res have been the cause of incalculable losses¼
Vast denuded areas, charred stubs and fallen trunks and
the general prevalence of blackened poles [illustrate] their
frequency and severity long before this country was
explored by white men¼Evidence indicates light ground
®res over practically the whole forest, some of the ®nest
stands of yellow pine show only slight charring of the
bark and very little damage to poles and undergrowth'
(Lang & Stewart, 1910, pp. 18±19). Dutton (1882, p.
136) observed that `the trees are large and noble in aspect
and stand widely apart, except in the highest parts of the
plateau where the spruces predominate. Instead of dense
thickets where we are shut in by impenetrable foliage, we
can look far beyond and see the tree trunks vanishing
away like an in®nite colonnade.' B. Vaughn, a cowboy
with the Grand Canyon Cattle Company, recalled being
able to `see a cow a half mile (0.8 km) in ponderosa
country (c. 1924), but now brush has grown up¼Fires
Figure 4 1909 photograph by G.A. Pearson:
`Tusayan National Forest ¼showing typical
group arrangement of age classes. South of
Grand View.' USDA Forest Service, Rocky
Mountain Research Station, Historical
Archives, Flagstaff, AZ.
Figure 5 1997 photograph in the Powell
Plateau (PP) study site. Sporadic surface ®res
occurred at Powell Plateau after 1879. The
site is characterized by groups of large trees,
productive understory communities, and
crown®re-resistant fuel complexes.
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
Natural variability in forests
started on the points and burned through needles and
small growth but didn't do much damage to larger trees¼
Indians burned areas near the rim when they were
gathering pin
Äon nuts to make the gathering easier' (Fossey,
1974). C. Wagner, a forester with the Park Service in
1935, described seeing through the pine forest for `three-
eighths of a mile (0.6 km)¼unobstructed. There was no
amount of undergrowth at all. At that time they had the
heavy deer population on the North Rim. No young oak
or aspen or anything' (Winchester, 1994).
The ponderosa pine age distribution on the North Rim
was unusually young for this species and region (Fig. 2).
Few trees predated 1650 and 95.5% of the pre-1880 tree
population was less than 300 year old-in 1998. The oldest
tree encountered had a centre date of 1537 (site PP). In
contrast, trees up to 540-year-old-were encountered at the
unharvested Gus Pearson Natural Area (Mast et al., 1999)
and at Mt Logan, about 70 km west of the North Rim sites
(Waltz & Fule
Â, 1998). Two hypotheses could explain the
short age distributions found on the North Rim: ®rst,
forests could have been affected by a stand-replacing
disturbance, such as severe ®re, before 1700. In central
Colorado, Brown et al. (1999) suggested that a `stand-
destroying event' could explain the relatively young trees on
an unharvested study site. However, the ®re history
reconstructed by Brown et al. (1999) included ®re-free gaps
up to 128 years long, allowing for increases in dead and
living fuels to support such a ®re. The North Rim ®re
histories, although temporally limited by the relatively
young ages of sampled trees, show steady, frequent ®res
from the early 1700s through 1879, with no indication of
extended gaps (Fule
Âet al., in review). A second hypothesis
is that North Rim ponderosa pines, growing on coarse-
textured limestone soils, may have shorter lifespans than
their counterparts on ®ne-textured basalt soils. Wind-
thrown trees were commonly observed on the study sites.
The early increase in populations of sprouting species,
coinciding with the exclusion of ®re, is consistent with the
continual reproductive capability of oaks and aspen. Under
the frequent ®re regime oaks may have maintained relatively
dense sprouts but few would survive repeated surface ®res.
Once ®res stopped, however, thickets of young ramets would
be well-positioned to become sapling and pole thickets. At
Camp Navajo, concurrent ponderosa pine harvesting may
have facilitated oak dominance by removing pine seed
sources (Fule
Âet al., 1997). But at the Grand Canyon sites,
pine harvesting occurred only after oak regeneration at KF
and early oak reproduction was extensive even at the
unharvested GV, PP, and RP sites.
Interacting factors of change
The spread of ponderosa pine forests across the Colorado
Plateau was associated with warming of climate since the
latest Wisconsin period, c. 11,000±14,000 years BP (Ander-
son, 1989; Weng & Jackson, 1999). Over much shorter time
periods, climatic ¯uctuation has been associated with tree
establishment in south-western ponderosa pine (Savage
et al., 1996). However, an analogous climate-regeneration
pattern was not observed in the present study. Grand
Canyon tree establishment dates were not concentrated in
the early twentieth century, in contrast to the pattern of high
regeneration in 1919 and other favourable moisture years
elsewhere in northern Arizona and the south-west (Schubert,
1974; Savage et al., 1996; Grissino-Mayer & Swetnam,
2000). Regeneration of sprouting oak and aspen species
began right after ®re exclusion and the seed-reproducing
species, ponderosa pine, white ®r and Douglas-®r, regener-
ated at a relatively consistent rate through most of the post-
settlement period (Fig. 2). Tree centre dates by species were
not signi®cantly correlated with the 20-year running average
of Palmer Drought Severity Index (PDSI) reconstructed from
dendrochronological data by Cook et al. (1996) for 1710±
1950. We compared all dated trees with those where the
estimated rings to pith were ®ve or less. There were
differences in the absolute numbers, because c. 50% of the
trees had an estimate >5 years to centre, but there was no
difference in the pattern.
In discussing the exceptional ponderosa pine regeneration
of 1919, Savage et al. (1996) recognized the novel circum-
stances of livestock grazing and ®re exclusion which
permitted dense seedling patches to ¯ourish, but they
speculated that the 1919 climate event might have caused
a signi®cant pine regeneration pulse even had these human-
caused disruptions not occurred. This hypothesis is not
supported by the Grand Canyon data: large increases in
seed-reproducing species occurred only at the ®re-excluded
sites (KF, GV and SR), not at the sites where at least a few
surface ®res occurred after 1879 (PP, RP and FP). Pine
density did increase signi®cantly at PP, but statistical
signi®cance may not equate with ecological signi®cance:
the magnitude of the increase was small (98.2 trees ha
65% increase) in comparison with increases in ®re-excluded
forests (hundreds to thousands of percent (Covington &
Moore, 1992; Covington et al., 1997; Fule
Âet al., 1997)).
These patterns suggest that recent climatic factors were less
in¯uential in affecting forest structure, especially as com-
pared with the human-caused factors described below.
Tree cutting
The Tusayan Ranger District (KF site) was one of the last
forests in northern Arizona south of the Colorado River to
be logged, because of its remoteness from the transconti-
nental railroad (Putt, 1995). The KF site was probably ®rst
logged between 1929 and 1931; a section of logging railroad
crossed the south-eastern corner of the study site (Putt, 1991
Stein, 1993). Within Grand Canyon National Park, old
stumps in the GV site date from a sawmill operated on the
Berry/Hearst (Grandview Hotel) property to supply mine
timbers (Sutphen, 1991). Trees were also cut in the Grand-
view site in dwarf mistletoe treatments (Lightle & Hawks-
worth, 1973), road and utility corridors, and `insect
treatments' on both rims, as well as cutting 1600 trees for
telephone poles on the South Rim and lumber for the lodge
at Bright Angel Point on the North Rim (Annual
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
12 P. Z. Fule
Âet al.
Superintendent's Reports, 1933 and 1936, on ®le at Grand
Canyon National Park). Pyne (1989) described snag cutting
during ®re suppression in the park.
Two implications arise from the tree cutting history. First,
the unharvested park sites retain a legacy of old-growth trees
that are missing from KF (Fig. 2). Second, tree cutting has
long lasting effects. Approximately 60 years after timber
harvest, pine density at KF was roughly equal to that of the
unharvested GV site, 541 vs. 646 trees ha
(Table 3). But
basal area at KF was only 62% that of GV, 14.1 vs.
22.9 m
, mostly because of a 6.5-m
difference in
pine basal area (Table 3).
Fire exclusion and herbivory
Domestic and wild mammalian herbivores affected forests in
at least three ways: disrupting frequent-®re regimes by
removing ®ne fuels, facilitating tree regeneration by elimin-
ating competing herbaceous plants, and direct herbivory of
young trees. Belsky & Blumenthal (1997) suggested that
focus on the effects of ®re suppression and logging in the
interior West has led to underestimation of the role of
livestock grazing in causing ecological degradation. No
understatement is evident in the south-west, where heavy
livestock grazing by early European settlers has been
consistently linked by numerous authors with the removal
of ®ne fuels and interruption of frequent ®re regimes
throughout the south-west (e.g. Leopold, 1924; Cooper,
1960; Swetnam & Baisan, 1996). Heavy livestock grazing
also coincided with ®re regime disruption at both earlier and
later dates than the typical 1870±1890 dates of European
settlement of the south-west, indicating that grazing itself
and not some coincidental factor was responsible for
initiating ®re exclusion. For example, grazing in the Rio
Grande valley had local effects on ®re regimes prior to 1800
(Baisan & Swetnam, 1997), Navajo sheep grazing led to ®re
cessation in north-eastern Arizona as early as 1830 (Savage
& Swetnam, 1990), and increased grazing in northern
Mexico resulted in ®re regime disruptions as late as 1930±50
Â& Covington, 1997, 1999).
In the Grand Canyon region, grazing of sheep and cattle
was heavy on both rims since European settlement until the
fencing of Park Service lands in the late 1930s. Ranches were
established at Short Creek and Pipe Spring, on the Arizona±
Utah border, as early as 1863 but the range was not
available for livestock until the settlement of hostilities with
Navajos in 1869 (Altschul & Fairley, 1989). The earliest
quanti®ed use of the `Kaibab mountain' and surrounding
areas was 1885±86, with about 2000 cattle; by 1887±89, use
of the west-side ranges increased to 200,000 sheep and
20,000 cattle (Rasmussen, 1941). By 1909, Forest Service
permits covered 14,000 head of cattle and horses, plus a
band of 5000 sheep on the Kaibab Plateau (Lang & Stewart,
1910). Mace (1990: p. 68) described `hundreds' of wild
horses ranging over the Plateau at that time. Near Grand-
view, Hull cabin was constructed in 1888 as the base for a
sheep ranching operation (Anderson, 1998).
Herbivory by wildlife increased substantially in the early
twentieth century. Following the designation of the Grand
Canyon Game Preserve in 1906, hunting was prohibited and
predator control efforts led to extirpation of the wolf and
massive killing of mountain lions, coyotes, and other
predators by 1931 (Rasmussen, 1941). The Kaibab deer
population irrupted after 1905, reaching an estimated
100,000 head of deer in 1924 before declining because of
starvation and government hunting programmes (Rasmus-
sen, 1941; Mitchell & Freeman, 1999
5). Adams (1925) noted
the `absence of young aspens caused mainly by overbrowsing
by deer and stock' on the North Rim (Adams, 1925, p. 589).
Near Grandview, he observed that the `condition of the
range in the forest and the park are equally bad, so that you
cannot tell by the appearance of the range whether you are in
the park or the forest' (Adams, 1925, p. 585).
Irrupting deer herds have reduced tree recruitment in
forests around the world: Japan (Abrams et al., 1999),
Patagonia (Veblen et al., 1989) and the eastern USA
(Abrams & Orwig, 1996). At Grand Canyon, mammalian
herbivory in¯uenced tree survival with long-term effects on
forest structure (Rasmussen, 1941; Merkle, 1954, 1962;
Mitchell & Freeman, 1993). Direct effects of excessive
herbivory appear to be most evident at the SR site.
Approximately 49 aspen trees ha
survived from establish-
ment in the ®rst decades after ®re exclusion (1880±1900),
but then a long demographic gap occurred between 1900
and 1960 when less than 5 aspen trees ha
(Fig. 2). Trampling and exposure of bare mineral soil by
deer and livestock may also have facilitated conifer regen-
eration (Mitchell & Freeman, 1993).
The consequences of elimination of approximately
60 years of aspen regeneration at SR may have far-reaching
effects. In the absence of aspen regeneration and ®re, white
®r and Douglas-®r dominated the post-settlement regener-
ation of the mesic site, leading to a conifer-dominated forest
that could support stand-replacing ®re because of the dense
forest ¯oor, high fuel loading, and low canopy base heights
(unpublished data). Intense burning occurred in the 1993
NWIII prescribed ®re, adjacent to the SR site (unpublished
data). Had excessive deer herbivory not occurred, the fuel
structures of an aspen/pine forest would probably be less
susceptible to crown®re and more conducive to long-term
ponderosa pine survival, although white ®r still may have
come to predominate in the absence of ®re.
Evidence of direct mammalian herbivory of tree species
other than aspen is unclear because there are no apparent
gaps in recruitment after ®re exclusion. Aspen is by far the
most palatable of the tree species on the study sites, but
Adams (1925) noted grazing on oak and cliffrose as well and
Rasmussen (1941, p. 254) photographed a ponderosa pine
seedling repeatedly browsed by deer. Insect herbivory may
also have played a role. Ponderosa pine defoliation by the
pandora moth (Coloradia pandora) has been shown to
interact with climate and ®re regime ¯uctuations in Oregon
(Speer, 1997). The Kaibab Plateau is the southernmost
region where large-scale pandora moth outbreaks have
occurred (Miller & Wagner, 1989).
Herbivory and ®re exclusion are usually so interlinked
that the relative importance of their separate in¯uences on
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
Natural variability in forests
increased tree recruitment is dif®cult to determine, but
comparisons have been made at a few rare study sites (Belsky
& Blumenthal, 1997). At Zion National Park (Utah) and on
Meeks Table (Washington), dense ungrazed herbaceous
plants appeared to have resisted the establishment of dense
ponderosa pine seedlings even in the absence of ®re
(Rummel, 1951; Madany & West, 1983). The Grand
Canyon study sites were all grazed by livestock through
the 1930s and all sites experienced high deer populations.
Although sprouting species established at all the study sites
since European settlement, density of seed-reproducing
species did not increase substantially at any of the sites with
post-settlement surface ®res (PP, FP and RP, Tables 3 and 6).
But density of seed-reproducing species did increase by as
much as several orders of magnitude at all the unburned sites
(KF, GV and SR, Tables 3 and 6). The differences in density
and species composition between burned and ®re-excluded
sites support the hypothesis that the thinning effect of
surface ®res was relatively more important than climate,
herbivory, or tree cutting, in regulating seedling or sprout
establishment and consequent forest density increases within
the park.
Are remote Grand Canyon forests `reference sites'?
A central theme of this paper is that sites PP, RP, and the
western third of FP, are currently in conditions similar to
those which prevailed prior to European settlement, so
contemporary characteristics of these sites can be used as
points of reference of natural variability. Even the ®re-
disrupted sites, eastern FP, SR and GV, are nonetheless
unharvested (`virgin') forests and provide useful compari-
sons. This information could be helpful for restoration and
conservation on the Kaibab Plateau ± and, to some extent,
broadly over the south-west and through the range of related
ecosystems. Is this valid?
There are two general arguments against placing much
reliance on concepts of `natural variability' or `reference
conditions.' First, these conditions may be poorly under-
stood or dif®cult to reconstruct (e.g. Stephenson, 1999;
Tiedemann et al., 2000). Second, even a good understanding
of conditions prior to European settlement may be of limited
relevance today given continual change: climate, extinctions,
invasions, management practices and evolution (e.g. Millar
& Wolfenden, 1999). With respect to the ®rst point, a
contemporary reference site can be measured for all varia-
bles of interest, so knowledge is not limited by the need to
reconstruct past conditions. However, clearly even the
`undisrupted' forests have changed in terms of ®re regime,
herbivory, invasion of exotic species, predator control,
native American resource management, etc. The only
appropriate option is to present the historical context as
completely as possible to permit an informed interpretation
of the data. With respect to the second point, we suggest that
reference information from contemporary sites is in fact
highly relevant. The concept of reference information is not
a static condition at a point in time, but rather an
understanding of the recent evolutionary environment of
an ecosystem (Moore et al., 1999). Contemporary reference
sites are unusually important because they show the variab-
ility in ecological conditions under today's climate, atmo-
spheric composition, and so on, corresponding most closely
to Stephenson's (1999) de®nition of `natural'.
Some have interpreted reference information as implicitly
prescriptive, translating directly into management goals (see
criticism of this approach in Tiedemann et al., 2000). Of all
resource management agencies, this may be most nearly true
for the National Park Service, mandated to manage for
`natural' conditions (Stephenson, 1999). In most settings,
though, reference information is applied to inform the design
of management strategies and selection of alternatives. The
natural range of variability may indeed be a useful manage-
ment goal, but it is rarely likely to be the only goal (Landres
et al., 1999).
Cole (2001) drew a helpful distinction between `wildness'
(minimal human manipulation) and `naturalness' (minimal
ecological degradation caused by `post-aboriginal human
in¯uence'). Grand Canyon forests are largely `wild' but fall
along a continuum of `naturalness.' The ®re-excluded forests
± especially SR ± exemplify the fact that unmanaged areas
are not necessarily natural. The National Park Service,
Forest Service, and other agencies are faced with contentious
management choices that may emphasize minimal manipu-
lation even if systems diverge greatly from the evolutionary
environment (wild/not natural), manipulation to restore
ecological function and structure (not wild/natural), or a
compromise position of intermittent and partial interven-
tion, such as sporadic prescribed burning. Cole (2001)
considered compromise the most likely outcome because of
the bureaucratic characteristics of resource management
agencies and the costs of restoration.
Simply through maintenance of ®re disturbance regimes
and protection against external degradation, the small but
important areas like the points and plateaus of the western
North Rim may retain key elements of both wildness and
naturalness, as far as possible given their size, atmospheric
composition change, and other factors. Remote relict
areas, which were least connected to pervasive changes in
disturbance regimes, appear to be currently in least
disrupted conditions. Now these sites provide valuable
opportunities to ask questions about dynamic or transitory
ecosystem elements that cannot be adequately studied
through retrospective or inferential means. Herbaceous
plant communities, wildlife habitat use, soil biogeochem-
istry and water relations, and invertebrate diversity are
examples of studies that can build on the `base datum' of
understanding of North Rim ®re regime and forest
We thank Grand Canyon National Park and Kaibab
National Forest staff assisting with this research, especially
R. Winfree, K. Kerr, D. Oltrogge, D. Spotskey, M. Schroe-
der, J. Schroeder, K. Crumbo, N. Brian, J. Balsam, A. Horn-
Wilson, R. Thakali, D. Snyder, D. Bertolette, and B. Higgins.
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
14 P. Z. Fule
Âet al.
Northern Arizona University's Ecological Restoration Insti-
tute's students and staff, especially J.P. Roccaforte, L.
Machina, L. Labate, B. Kent, M. Wilkinson Kaye,
M. Stoddard, J. Crouse, S, Curren, and H.B. `Doc' Smith,
supported the study. V and K. Zarlingo provided ®eld
support. Several anonymous reviewers provided helpful
comments on the manuscript. This work was funded by a
grant from the U.S. Department of the Interior.
Abrams, M.D., Copenheaver, C.A., Terazawa, K., Umeki, K.,
Takiya, M. & Akashi, N. (1999) A 370-year dendroecolog-
ical history of an old-growth Abies-Acer-Quercus forest in
Hokkaido, northern Japan. Canadian Journal of Forest
Research,29, 1891±1899.
Abrams, M.D. & Orwig, D.A. (1996) A 300-year history of
disturbance and canopy recruitment for co-occurring white
pine and hemlock on the Allegheny Plateau, USA. Journal of
Ecology,84, 353±363.
Adams, C.C. (1925) Ecological conditions in National Forests
and in National Parks. Scienti®c Monthly,20, 561±593.
Altschul, J.H. & Fairley, H.C. (1989) Man, models, and
management: an overview of the archaeology of the Arizona
Strip and the management of its cultural resources. USDA
Forest Service and USDI Bureau of Land Management Report
contract # 53-8371-6-0054), submitted by Dames & Moore
Anderson, R.S. (1989) Development of the southwestern pon-
derosa pine forests: what do we really know? Multiresource
management of Ponderosa pine forests (eds A. Tecle, W.W.
Covington and R.H. Hamre) (Technical coords.), pp. 15±22.
USDA Forest Service General Technical Report RM-185,
Rocky Mountain Forest and Range Experiment Station, Fort
Collins, CO.
Anderson, M.F. (1998) Living at the edge: explorers, exploiters,
and settlers of the Grand Canyon Region. Grand Canyon
Association, AZ.
Anderson, M.K. & Moratto, M.J. (1996) Native American
land-use practices and ecological impacts. Sierra Nevada
ecosystem project: ®nal report to congress, Vol II
7, pp. 187±
206. University of California, Davis, CA.
Applequist, M.B. (1958) A simple pith locator for use with off-
center increment cores. Journal of Forestry,56, 141.
Baisan, C.H. & Swetnam, T.W. (1997) Interactions of ®re
regimes and land use in the central Rio Grande valley.
USDA Forest Service Research Paper RM-RP-330, Rocky
Mountain Forest and Range Experiment Station, Fort
Collins, CO.
Barger, R.L. & Ffolliott, P.F. (1972) Physical characteristics and
utilization of major woodland tree species in Arizona. USDA
Forest Service Research Paper RM-83, Rocky Mountain
Forest and Range Experiment Station, Fort Collins, CO.
Belsky, A.J. & Blumenthal, D.M. (1997) Effects of livestock
grazing on stand dynamics and soils in upland forests of the
Interior West. Conservation Biology,11, 315±327.
Bennett, P.S. (1974) The ®re ecology of the ponderosa pine
ecosystem of the North Rim forests, Grand Canyon National
Park. Report on ®le at Grand Canyon National Park, Grand
Canyon National Park, AZ.
Brown, P.M., Kaufmann, M.R. & Shepperd, W.D. (1999) Long-
term, landscape patterns of past ®re events in a montane
ponderosa pine forest of central Colorado. Landscape Ecol-
ogy,14, 513±532.
Cole, D.N. (2001) Management dilemmas that will shape
Wilderness in the 21st century. Journal of Forestry,99, 4±8.
Cook, E.R., Meko, D.M., Stahle, D.W. & Cleaveland, M.K.
(1996) Tree-ring reconstructions of past drought across the
conterminous United States: tests of a regression method and
calibration/veri®cation results. Tree rings, environment, and
humanity: proceedings of the international conference (eds
J.S. Dean, D.M. Meko and T.W. Swetnam), pp. 155±170.
Radiocarbon: Department of Geosciences, The University of
Arizona, Tucson, AZ, USA.
Cooper, C.F. (1960) Changes in vegetation, structure, and
growth of southwestern pine forests since white settlement.
Ecology,42, 493±499.
Covington, W.W. & Moore, M.M. (1992) Post-settlement
changes in natural disturbance regimes: implications for
restoration of old-growth ponderosa pine ecosystems. Old-
growth forests in the southwest and Rocky Mountain region.
Proceedings of the symposium, pp. 81±99. USDA Forest
Service General Technical Report RM-213. Rocky Mountain
Forest and Range Experiment Station, Fort Collins, CO.
Covington, W.W. & Moore, M.M. (1994) Southwestern pon-
derosa forest structure and resource conditions: changes
since Euro-American settlement. Journal of Forestry,92,
Covington, W.W., Fule
Â, P.Z., Moore, M.M., Hart, S.C., Kolb,
T.E., Mast, J.N., Sackett, S.S. & Wagner, M.R. (1997)
Restoration of ecosystem health in southwestern ponderosa
pine forests. Journal of Forestry,95, 23±29.
Denevan, W.M. (1992) The pristine myth: the landscape of the
Americas in 1492. Annals of the Association of American
Geographers,82, 369±385.
Duhnkrack, J.A. (1982) Fire regime analysis of the ponderosa
pine forest at the Grandview Monocline on the South Rim of
Grand Canyon National Park, Arizona. Undergraduate
Report, Huxley College of Environmental Studies. Western
Washington University, Bellingham, WA.
Dutton, C.E. (1882) Tertiary history of the Grand Canyon. U.S.
Geological Survey Monograph II, Washington, DC.
Fossey, W.R. (1974) Oral interview of Aldus `Blondie' Jensen
and Bob Vaughn: personal interview, Fredonia, Arizona,
September 6, 1974. On ®le at Grand Canyon National Park,
Grand Canyon National Park, AZ.
Â, P.Z. & Covington, W.W. (1997) Fire regimes and forest
structure in the Sierra Madre Occidental, Durango, Mexico.
Acta Botanica Mexicana,41, 43±79.
Â, P.Z. & Covington, W.W. (1999) Fire regime changes in La
Michila Biosphere Reserve, Durango, Mexico. Conservation
Biology,13, 640±652.
Â, P.Z., Heinlein, T.A., Covington, W.W. & Moore, M.M.
Fire regimes on an environmental gradient in Grand Canyon
National Park, Arizona, USA. Ecological Applications (in
Â, P.Z., Moore, M.M. & Covington, W.W. (1997) Deter-
mining reference conditions for ecosystem management in
southwestern ponderosa pine forests. Ecological Applica-
tions,7, 895±908.
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
Natural variability in forests
Ganey, J.L. & Block, W.M. (1994) A comparison of two
techniques for measuring canopy closure. Western Journal of
Applied Forestry,9, 21±23.
Ganey, J.L., Block, W.M., Jenness, J.S. & Wilson, R.A. (1999)
Mexican spotted owl home range and habitat use in pine-oak
forest: implications for forest management. Forest Science,
45, 127±135.
GCNP, [Grand Canyon National Park] (1992) Fire management
plan. On ®le at Grand Canyon National Park, Grand Canyon
National Park, AZ.
Gottfried, G.J. (1980) Control of New Mexican locust and
the effect on planted ponderosa pine in central Arizona.
USDA Forest Service Research Note RM-386, Rocky
Mountain Forest and Range Experiment Station, Fort
Collins, CO.
Grissino-Mayer, H.D. & Swetnam (1997) Multi-century history
of wild®re in the ponderosa pine forests of El Malpais.
Natural history of El Malpais National Monument (ed.
K. Mabery). New Mexico Bureau of Mines and Mineral
Resources Bulletin 156, 163±171.
Grissino-Mayer, H.D. & Swetnam, T.W. (2000) Century-scale
climate forcing of ®re regimes in the American Southwest.
Holocene,10, 213±220.
Huffman, D.W., Moore, M.M., Covington, W.W., Crouse, J.E.
& Fule
Â, P.Z. Ponderosa pine forest reconstruction: compar-
isons with historical data (peer-reviewed). Proceedings: steps
toward stewardship. Ponderosa pine ecosystem restoration
and conservation. Proceedings of the RMRS-P-000. Ogden,
UT: U.S. Department of Agriculture, Forest Serice, Rocky
Mountain Research Station, UT.
Laacke, R.J. (1990) Abies Concolor (Gord. and Glend.) Lindl.
Ex Hildebr. White Fir. Silvics of North America: 1. conifers
(eds R.M. Burns and B.H. Honkala) (Technical Coords.), pp.
36±46. USDA Forest Service Agriculture Handbook 654,
USDA Forest Service, Washington, DC.
Landres, P., Morgan, P. & Swanson, F. (1999) Overview of the
use of natural variability in managing ecological systems.
Ecological Applications,9, 1279±1288.
Lang, D.M. & Stewart, S.S. (1910) Reconnaissance of the Kaibab
National Forest. Unpublished report on ®le at Northern
Arizona University, Arizona University, Flagstaff, AZ.
Leopold, A. (1924) Grass, brush, timber, and ®re in southern
Arizona. Journal of Forestry,22, 1±10.
Leopold, A. (1937) Conservationist in Mexico. American
Forests,37, 118±120, 146.
Leopold, A. (1941) Wilderness as a land laboratory. Living
Lightle, P.C. & Hawksworth, F.G. (1973) Control of dwarf
mistletoe in a heavily used ponderosa pine recreation forest:
Grand Canyon, Arizona. USDA Forest Service Research
Paper RM-106, Rocky Mountain Forest and Range Experi-
ment Station, Fort Collins, CO.
Mace, W. (1990) Will Mace Memoirs. People and places of
the old Kaibab (eds C.P. Moose, T.A. Cleeland, P.J. Putt
and W. Mace), pp. 65±73. USDA Forest Service Cultural
Resources Management Report no. 10, Southwestern Region,
USDA Forest Service, Albuquerque, NM.
Madany, M.H. & West, N.E. (1983) Livestock grazing ± ®re
regime interactions within montane forests of Zion National
Park. Ecology,64, 661±667.
Maser, C., Anderson, R.G., Cromack, K., Williams, J.T. Jr &
Martin, R.E. (1979) Dead and down woody material.
Wildlife habitats in managed forests ± the Blue Mountains
of Oregon and Washington, pp. 78±95. USDA Agricultural
Handbook 553, USDA, Washington, DC.
Mast, J.N., Fule
Â, P.Z., Moore, M.M., Covington, W.W. &
Waltz, A. (1999) Restoration of presettlement age structure of
an Arizona ponderosa pine forest. Ecological Applications,9,
Merkle, J. (1954) An analysis of the spruce-®r community on the
Kaibab Plateau, Arizona. Ecology,35, 316±322.
Merkle, J. (1962) Plant communities of the Grand Canyon area,
Arizona. Ecology,43, 698±711.
Millar, C.I. & Wolfenden, W. (1999) The role of climate change
in interpreting historical variability. Ecological Applications,
9, 1207±1216.
Miller, K.K. & Wagner, M.R. (1989) Effect of pandora moth
(Lepidoptera: Saturniidae) defoliation on growth of ponder-
osa pine in Arizona. Journal of Economic Entomology,82,
Minnich, R.A., Barbour, M.G., Burk, J.H. & Fernau, R.F. (1995)
Sixty years of change in Californian conifer forests of the
San Bernadino Mountains. Conservation Biology,9, 902±914.
Mitchell, J.E. & Freeman, D.R. (1993) Wildlife-lifestock±®re
interactions on the North Kaibab: a historical review. USDA
Forest Service General Technical Report RM-222, Rocky
Mountain Forest and Range Experiment Station, Fort Collins,
Moore, M.M., Covington, W.W. & Fule
Â, P.Z. (1999) Evolu-
tionary environment, reference conditions, and ecological
restoration: a southwestern ponderosa pine perspective.
Ecological Applications,9, 1266±1277.
NPS, [National Park Service] (1992) Western Region ®re
monitoring handbook. NPS, San Francisco, CA.
Putt, P.J. (1995) South Kaibab National Forest: a historical
overview to 1940. MA Thesis. Northern Arizona University,
Flagstaff, AZ.
Pyne, S.J. (1989) Fire on the Rim. Wieden®eld & Nicholson,
New York.
Rasmussen, D.I. (1941) Biotic communities of Kaibab Plateau,
Arizona. Ecological Monographs 11, 229±275.
Reeberg, P. (1995) The western region ®re monitoring hand-
book. USDA forest service general technical report INT-
GTR-320, pp. 259±260. Intermountain Forest and Range
Experiment Station, Ogden, UT.
Rummel, R.S. (1951) Some effects of livestock grazing on
ponderosa pine forest and range in central Washington.
Ecology,32, 594±607.
Savage, M., Brown, P.M. & Feddema, J. (1996) The role of
climate in a pine forest regeneration pulse in the southwestern
United States. Ecoscience,3, 310±318.
Savage, M. & Swetnam, T.W. (1990) Early 19th century ®re
decline following sheep pasturing in a Navajo ponderosa pine
forest. Ecology,71, 2374±2378.
Schubert, G.H. (1974) Silviculture of southwestern ponderosa
pine: the status-of-our-knowledge. USDA Forest Service
Research Paper RM-123, Rocky Mountain Forest and Range
Experiment Station, Fort Collins, CO.
Sellers, W.D. & Hill, R.H. (1974) Arizona Climate 1931±72.
The University of Arizona Press, Tucson, AZ.
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
16 P. Z. Fule
Âet al.
Speer, J.H. (1997) A dendrochronological record of pandora
moth (Coloradia pandora, Blake) outbreaks in central Ore-
gon. MSc Thesis. Department of Geosciences, University of
Arizona, Tucson, AZ.
Stein, P. (1993) Logging Railroads of the Coconino and Kaibab
National Forests supplemental report to a National Register
of Historic Places documentation form. SWCA Archeology
Report no. 93±16, on ®le at Kaibab National Forest, Kaibab
National Forest, Williams, AZ.
Stephenson, N.L. (1999) Reference conditions for giant sequoia
forest restoration: structure, process, and precision. Ecolog-
ical Applications,9, 1253±1265.
Stokes, M.A. & Smiley, T.L. (1968) An introduction to tree-ring
dating. University of Chicago Press, Chicago.
Sutphen, D.L. (1991) Grandview, Hermit, and South Kaibab
trails: linking the past, present and future at the Grand
Canyon of the Colorado. MA Thesis. Northern Arizona
University, Flagstaff, AZ.
Swetnam, T.W., Allen, C.D. & Betancourt, J.L. (1999) Applied
historical ecology: using the past to manage for the future.
Ecological Applications,9, 1189±1206.
Swetnam, T.W. & Baisan (1996) Historical ®re regime patterns
in the southwestern United States since AD 1700. Proceedings
of the 2nd la Mesa Fire Symposium (ed. C.D. Allen), pp. 11±
32.USDA Forest Service General Technical Report RM-
GTR-286, Rocky Mountain Forest and Range Experiment
Station, Fort Collins, CO.
Swetnam, T.W. & Betancourt, J.L. (1998) Mesoscale disturbance
and ecological response to decadal climatic variability in the
American Southwest. Journal of Climate,11, 3128±3147.
Thomas, J.W., Anderson, R.G., Maser C. & Bull E.L. (1979)
Snags. Wildlife habitats in managed forests ± the Blue
Mountains of Oregon and Washington, pp. 60±77. USDA
Agricultural Handbook, 553, USDA, Washington, DC.
Tiedemann, A.R., Klemmedson, J.O. & Bull, E.L. (2000)
Solution of forest health problems with prescribed ®re: are
forest productivity and wildlife at risk? Forest Ecology and
Management,127, 1±18.
Veblen, T.T., Mermoz, M., Martin, C. & Ramilo, E. (1989)
Effects of exotic deer on forest regeneration and composition in
northern Patagonia. Journal of Applied Ecology,26, 711±724.
Verkamp, M.M. (1940) History of Grand Canyon National
Park, MA Thesis. Department of History, University of
Arizona, Tucson, AZ (reprinted by Grand Canyon Pioneers
Society, Flagstaff, AZ).
Waltz, A.E.M. & Fule
Â, P.Z. (1998) Changes in ponderosa pine
forests of the Mt. Logan Wilderness. On File at Northern
Arizona University, Northern Arizona University, Flagstaff,
Warren, P.L., Reichardt, K.L., Mouat, D.A., Brown, B.T. &
Johnson, R.R. (1982) Technical report no. 9. Vegetation of
Grand Canyon National Park. Contracts no. CX8210-7±
0028 and CX8000-9±0033, Contribution no. 017/06. On ®le
at Grand Canyon National Park, National Park Service/
University of Arizona, AZ.
Weaver, H. (1951) Fire as an ecological factor in the southwest-
ern ponderosa pine forests. Journal of Forestry,49, 93±98.
Weng, C. & Jackson, S.T. (1999) Late-glacial and Holocene
vegetation history and paleoclimate of the Kaibab Plateau,
Arizona. Paleogeography, Paleoclimatology, Paleoecology,
153, 179.
White, A.S. (1985) Presettlement regeneration patterns in a
southwestern ponderosa pine stand. Ecology,66, 589±594.
White, M.A. & Vankat, J.L. (1993) Middle and high elevation
coniferous forest communities of the North Rim region of
Grand Canyon National Park, Arizona, USA. Vegetatio,109,
Winchester, J. (1994) Oral history interview with Claude `Bru'
Wagner, 75th anniversary celebration events at Grand Canyon
National Park. Unpublished document on ®le at Grand
Canyon National Park, Grand Canyon National Park, AZ.
Wolf, J.J. & Mast, J.N. (1998) Fire history of mixed-conifer
forests on the North Rim, Grand Canyon National Park,
Arizona. Physical Geographyraphy,19, 1±14.
Woolsey, T.S. Jr (1911) Western Yellow Pine in Arizona and
New Mexico. USDA Forest Service Bulletin 101, Government
Printing Of®ce, Washington, DC, 64 p.
Authors are Assistant Professor, Regents' Professor,
Professor, Senior Research Specialist, and Research Spe-
cialist. Research interests include ecological restoration of
®re-adapted forests, plant community ecology, ®re regime
studies, and invertebrate ecology.
ÓBlackwell Science Ltd 2002, Journal of Biogeography,29, 1±17
Natural variability in forests
Journal: Journal of Biogeography
Article: 655
Dear Author,
During the preparation of your manuscript for publication, the queries listed below have arisen.
Please could you respond to these queries by making the necessary corrections DIRECTLY ON
THE PROOF. Please ONLY USE the ‘Remarks’ column for clarification or comments. When
adding your corrections to the proof, please WRITE CLEARLY, in a strong blue or black ink, not
in capitals unless these are intended, and not too close to the paper edge if returning the proof by
fax. Please help us to publish your article accurately and quickly by following these guidelines.
Many thanks for your assistance.
Query Refs. Query Remarks
1 Au: Please approve the suggested short title.
2 Altschul & Fairly 1989 has been changed to
Altschul & Fairley 1989 so that this citation
matches the list
3 Huffman et al. 2001 has not been included in the
4 Putt 1991 has not been included in the list
5 Mitchell & Freeman 1999 has not been included
in the list
6 Au: Please provide publisher’s name and location.
7 Au: Please provide name/s of editor/s.
8 Au: Please provide editor/s name/s, if applicable.
9 Fulé et al. in press has not been found in the text
10 Au: Please update Fulé et al. (in press) with year
of publication, volume no. and page range.
11 Huffman et al. in press has not been found in the
12 Au: Please provide name/s of editor/s, if
13 Au: Please provide name/s of editor/s, if
14 Au: Please provide name/s of editor/s, if
... A common finding in dendroecological studies is that warm, dry conifer forests developed in uneven-aged pulses of regeneration and mortality and often not as a single-aged cohort, both tracking favorable climatic periods and slightly longer fire free intervals ). These uneven-aged distributions in tree ages have been documented in dry, mixed-conifer and ponderosa pine forests throughout the intermountain west (Mast and Veblen 1999;Fulé et al. 2002;Boyden et al. 2005;Brown and Wu 2005). ...
... Given the diversity of sites and ecosystems, it is not surprising that recent records of post-fire regeneration dynamics in montane and subalpine mixed-conifer forests report variable findings. Meanwhile, multiple other studies have identified large peaks of tree establishment since European settlement and fire suppression resulting in long fire-free periods (Fulé et al. 2002;Mast and Wolf 2004;Meunier et al. 2014). For example, Mast and Wolf (2004) reported a large establishment of trees between 1900 and 1950 that corresponded to favorable climate, livestock grazing, and active fire suppression, but prior to this period large pulses of establishment were missing from the record as far back as the 1720s. ...
Wildfire-mediated changes to forests have prompted numerous studies on post-fire forest recovery of coniferous forests. Given climate change, a growing body of work demonstrates that conifer regeneration in temperate and boreal forests is declining, a phenomenon often termed “regeneration failure.” However, the definition and parameters are numerous and variable. Characterization of drought also varies greatly, thus hindering the ability to compare results among areas. This review discusses new perspectives on conifer regeneration failure and places these studies into the context of drought and fire activity. We focus this review on three forest types where conifer regeneration failure is well documented: western boreal forests, cold mixed-conifer, and dry pine forests. To place the challenges to conifer tree regeneration in the context of regional climate trends, we present a novel regional analysis that summarizes drought conditions prior, during, and following the year of a large wildfire. We demonstrate the need to assess failure in the context of specific forest dynamics and well-defined metrics. For example, tree establishment may historically occur over longer periods, and current and future climate may exacerbate this and not promote pre-fire forest structure and composition. Many forests are undergoing rapid change and the type, magnitude, and causes of changes need to be compared among areas. Thus, we should be cautious of quantifying “regeneration failure” and drought without providing spatial and temporal context.
... Wildfire is a critical ecological process in ecosystems throughout the world [72][73][74][75]. Fire shapes and maintains the diversity, structure, and function of vegetation communities, even in areas where it has historically occurred infrequently [72,76]. The characteristics of many fire regimes have been influenced by unprecedented anthropogenic disturbances over the past few centuries [10,15,76]. ...
... Fire shapes and maintains the diversity, structure, and function of vegetation communities, even in areas where it has historically occurred infrequently [72,76]. The characteristics of many fire regimes have been influenced by unprecedented anthropogenic disturbances over the past few centuries [10,15,76]. These disturbances include over a century of fire suppression that has effectively removed the regulating force of wildfire [77]. ...
Full-text available
As the effects of climate change accumulate and intensify, resource managers juggle existing goals and new mandates to operationalize adaptation. Fire managers contend with the direct effects of climate change on resources in addition to climate-induced disruptions to fire regimes and subsequent ecosystem effects. In systems stressed by warming and drying, increased fire activity amplifies the pace of change and scale of severe disturbance events, heightening the urgency for management action. Fire managers are asked to integrate information on climate impacts with their professional expertise to determine how to achieve management objectives in a changing climate with altered fire regimes. This is a difficult task, and managers need support as they incorporate climate adaptation into planning and operations. We present a list of adaptation strategies and approaches specific to fire and climate based on co-produced knowledge from a science–management partnership and pilot-tested in a two-day workshop with natural resource managers and regional stakeholders. This “menu” is a flexible and useful tool for fire managers who need to connect the dots between fire ecology, climate science, adaptation intent, and management implementation. It was created and tested as part of an adaptation framework used widely across the United States and should be applicable and useful in many fire-prone forest ecosystems.
... In both, fire and non-fire ecosystems, fires can lead to the loss of habitats and local or global species extinction (Longán et al., 1999;Dennis et al., 2001;Nasi et al., 2002Nasi et al., , 2009) and can change species composition and forest structure (e.g. Cameeraat and Imeson, 1999;DeBano, 2000;Fulé et al., 2002;Türkmen and Düzenli, 2005) and their microclimate (Ma et al., 2010). For that reason, increasing knowledge on the responses of plant communities to disturbances such as fire is a key step and a necessary tool for adopting land management strategies that aim at effective biodiversity conservation (Seastedt et al., 2008). ...
Full-text available
In non-fire prone ecosystems, like some subtropical humid forests, fire produces habitat destruction and intensifies land degradation by inducing changes in native species composition, soil properties and erosive processes. Bryophytes are key components of the Macaronesian laurel forests playing an important role in regulating water cycling and microclimate. Ecological and taxonomical bryophytes groups have distinct ecological and physiological requirements and may respond in a different way to the same fire events. Therefore, analysing post fire recovery of bryophyte communities represents a key step towards a better understanding of forest fire drivers and post fire management. We investigated how species richness and composition of different ecological and taxonomical bryophyte groups varied in 1158 samples within a fire chronosequence from 5 to 57 years in the best-preserved laurel forest from Canary Islands (Garajonay National Park) analysing communities in terms of differences with comparable surrounding old growth unburnt stands. Epiphyte, terricolous and saxicolous bryophytes were sampled at each plot and the influence of the time since fire was analyzed together with environmental variables (temperature, precipitation, mist precipitation and elevation) and forest structure variables. Our results indicate that there is no general pattern of post fire recolonization, as recolonization varies depending on the ecological and phylogenetic groups considered. Climate and forest structure play an important role in post-fire recolonization, such that time since fire is not the most important variable influencing richness and composition. The results increase the understanding of the processes that shape compositional patterns in groups with high dispersal capacities and high microclimate dependence, such as mosses and liverworts.
... Fires in the southwestern USA were actively suppressed by public land management agencies during the majority of the twentieth century, leading to an increase in fuel loads and fuel connectivity and favoring the expansion of species not adapted to fire (Fulé et al. 2002;Margolis et al. 2007;Mast and Wolf 2004;Strahan et al. 2016). In mixed-conifer forests, fine surface fuels are typically sparse and ladder fuels more abundant than in ponderosa pine forests, which historically caused more infrequent and larger mixed-severity fires (Schoennagel et al. 2004). ...
Full-text available
Over the twenty-first century, the combined effects of increased fire activity and climate change are expected to alter forest composition and structure in many ecosystems by changing postfire successional trajectories and recovery. Southwestern US mountain ecosystems contain a variety of vegetation communities organized along an elevation gradient that will respond uniquely to changes in climate and fire regime. Moreover, the twentieth-century fire exclusion has altered forest structure and fuel loads compared to their natural states (i.e., without fire suppression). Consequently, uncertainties persist about future vegetation shifts along the elevation gradient. In this study, we simulated future vegetation dynamics along an elevation gradient in the southwestern US comprising pinyon-juniper woodlands, ponderosa pine forests, and mixed-conifer forests for the period 2000-2099, to quantify the effects of future climate conditions and projected wildfires on species productivity and distribution. While we expected to find larger changes in aboveground biomass, species diversity and species-specific abundance at low elevation due to warmer and drier conditions, the largest changes occurred at high elevation in mixed-conifer forests and were caused by wildfire. The largest increase in high-severity and large fires were recorded in this vegetation type, leading to high mortality of the dominant species, Picea engelmannii and Abies lasiocarpa, which are not adapted to fire. The decline of these two species reduced biomass productivity at high elevation. In ponderosa pine forests and pinyon-juniper woodlands, fewer vegetation changes occurred due to higher abundance of well-adapted species to fire and the lower fuel loads mitigating projected fire activity, respectively. Thus, future research should prioritize understanding of the processes involved in future vegetation shifts in mixed-conifer forests in order to mitigate both loss of diversity specific to high-elevation forests and the decrease in biomass productivity, and thus carbon storage capacity, of these ecosystems due to wildfires.
Full-text available
Interest in use of naturally ignited wildland fires managed to meet multiple resource objectives (resource objective wildfire) is increasing among U.S. public forest managers; however, only a limited number of studies have examined this approach for conserving or restoring understory plant diversity, productivity, and community structure. We analyzed understory community changes two and twelve years after resource objective fire, using permanent sample plots in three montane contrasting forest types in Grand Canyon National Park, AZ. Our findings indicated that species composition in the pine-oak forest rebounded to be similar to that observed before the fire, but plant cover did not recover to pre-fire levels by the twelfth year post-fire. Plant cover showed mixed results post-fire in mixed-conifer and spruce-fir forests, and species composition was still characterized by ruderal species twelve years later. Patterns observed in this study likely reflect interacting factors of burn severity, periodic drought, large ungulate herbivory, and inherent site variability. Other than cheatgrass (Bromus tectorum), we found no non-native species occurring with high frequency. Continued monitoring with increased frequency and intensity could lead to better understanding of long-term changes in these forests after resource objective fire, and enhance our understanding of important drivers of variation including interactions of climate, burn severity and herbivory by large ungulates.
Full-text available
Context Managers aiming to utilize wildland fire to restore southwestern ponderosa pine landscapes require better understanding of forest cover patterns produced at multiple scales. Restoration effectiveness of wildland fires managed for resource benefit can be evaluated against natural ranges of variation. Objectives We describe landscape patterns within reference landscapes, including restored and functioning ponderosa pine forests of northern Arizona, and compare them to wildland fires managed for resource benefit. We make comparisons along a gradient of extents and assess the effects of scale on landscape differences. Methods Using Sentinel-2 imagery, we classified ponderosa pine forest cover and calculated landscape metrics across a gradient of landscape extent within reference and managed landscapes. We used non-parametric tests to assess differences. We used random forest models to assess and explore which landscape metrics were most importance in differentiating landscape patterns. Results Restored forests exhibited landscapes patterns consistent with those of ecologically intact forest landscapes. Managed wildfire landscape patterns differed significantly when compared to reference landscape patterns among nearly all landscape metrics considered and became increasingly different with increasing landscape extent (15–840 ha), tending towards both denser and larger patch areas. Conclusions Landscape patterns from wildland fires managed for resource benefit we examined differ from those of reference landscapes. Differences become more pronounced with increasing landscape size. Landscape patterns among large managed forest landscapes suggest that the predominately single-entry, low-severity disturbance regime from these managed fires is failing to reduce tree densities and break up large contiguous areas of canopy cover.
Full-text available
Understanding naturally occurring pine regeneration dynamics in response to thinning and burning treatments is necessary not only to measure the longevity of the restoration or fuels treatment, but also to assess how well regeneration meets forest sustainability guidelines and whether natural regeneration is sufficient for maintaining a sustainable forest structure and composition. A synthesis review was carried out on the effects of mechanical thinning and prescribed burn treatments on natural pine regeneration response in frequent-fire ponderosa pine forests across the western United States. The focus was on site-specific variability in pine regeneration dynamics, temporal trends in regeneration presence and abundance, and response to treatment as described in the current literature using 29 studies that met our evidence-based review protocols. Data showed that the effects of thinning and burning treatments on regeneration depended on time since treatment. Mechanical thinning, prescribed burning, and thinning plus burn treatments all increased seedling density, but there was high variability among sites and studies. There were mixed results in the short-term (< 10 years) with both increasing and decreasing regeneration, and a general increase in regeneration 11 − 20 years post-treatment. Some long-term studies (> 20 years) concluded that stands can return to pre-treatment densities in terms of total trees per hectare and forest floor duff levels when there are no maintenance treatments applied. Several studies showed the average ponderosa pine seedling presence, survival and growth found in today’s forests to be at a high density; this combined with missed fire cycles could contribute to future fire risk and reduce the efficacy of maintaining fuel reduction goals.
Broad-scale forest restoration projects are implemented across the western United States to restore seasonally dry, frequent-fire-adapted ecosystems to improve ecological function and enhance resilience by increasing resistance to crown fire and climatic stressors. Despite the widespread use of restoration treatments that center on tree thinning and application of prescribed fire, the longevity of beneficial effects and the robustness of outcomes under future climate change predictions remains unclear. In this study, we remeasured a set of experimental areas established a minimum of 20 years ago that comprise a network of ponderosa pine (Pinus ponderosa) forest restoration study sites in northern Arizona. We analyzed ecological resiliency by evaluating forest conditions in terms of resistance to climatic stressors and potential crown fire in units that were thinned following evidence-based restoration guidelines (ERG), then burned with prescribed fire at multiple intervals, compared against paired untreated controls. Resilience indicators included forest structure, tree mortality, tree growth, regeneration, canopy fuels, and crowning index. We also simulated future forest conditions under a warming climate scenario (RCP 4.5) with a range of prescribed fire return intervals. Results indicated that experimental areas where restoration treatments were implemented remained more resilient to climate stressors compared to controls after 20 years. Treated areas had significantly lower tree mortality and greater average diameter growth compared to controls. Furthermore, forest structure generally remained similar to historical reference conditions in treated units with the exception of increases in ingrowth of sprouting species at the drier sites. Canopy fuel load and crown fire hazard in treated units remained significantly lower than controls, indicating that treatments remained effective in reducing crown fire potential over the 20-year study period without the need for additional tree thinning. Modeling basal area, crowning index, and the proportion of basal area in large trees under a future warming scenario suggested that the treated units underwent less changed than untreated areas. Under climate change, management of fire regimes even at longer-than-historical intervals (historical ≈ 5 yr, tested 5, 10, and 20 yr) would maintain basal area within our historical range of variability and maintain fire resistant forest over the next several decades. However, decline by the end of the century is concerning. Our results suggest that forest restoration treatments, guided by historical reference conditions, promote ecological resilience in the long-term and continued maintenance burning into the future is likely warranted even with continued drought and warming.
Ecological restoration of fire-excluded yellow pine-mixed conifer (YPMC) forests in the western US first requires the definition of desired future conditions. The Sierra de San Pedro Martír (SSPM) in Baja California has been used as a modern reference for other western US YPMC forests because it shares similar vegetation, soils, climate, and fire regimes, but is less impacted by logging and fire suppression. Our study re-visited a 14 ha SSPM watershed 16 years after it experienced a moderate severity fire preceded by four years of severe drought, in order to better understand resilient forests. We also provide a management-applicable metric for reference conditions by characterizing density and inter-tree competition using stand density index (SDI). We found no significant differences in forest structure between 2004 (1-yr post fire) and 2019 (16-yr post fire) periods, except for moderate and highly variable increases in duff and 1000-hr rotten fuel loads. Our exploration of SDI suggests that this watershed exists at extremely low levels of site occupancy and inter-tree competition relative to its potential. We concluded that, more than competition for resources, fire activity and large-scale climate fluctuations are the most important drivers of stand structure in the SSPM. This study demonstrates the importance of managing for very low tree densities as a component of resilience in YPMC forests.
Forest managers of the western United States are increasingly interested in utilising naturally ignited wildfires to achieve management objectives. Wildfires can accomplish a range of objectives, from maintenance of intact ecological conditions, to ecosystem restoration, to playing vital natural disturbance roles; however, few studies have carefully evaluated long-term effectiveness and outcomes of wildfire applications across multiple forest types. We remeasured monitoring plots more than 10 years after ‘resource objective’ (RO) fires were allowed to burn in three main south-western forest types. Results showed minimal effects and effective maintenance of open conditions in an intact pine-oak site. Higher-severity fire and delayed mortality of larger and older trees contributed to reductions in basal area and canopy cover at the mixed-conifer and spruce-fir sites. Species dominance shifted towards ponderosa pine in both the mixed-conifer and spruce-fir sites. Although fires resulted in 46–68% mortality of smaller trees initially, substantial ingrowth brought tree density to near pre-fire levels in all forest types after 12 years. Overall, the 2003 RO fires were broadly successful at maintaining or creating open and heterogeneous conditions and resulted in fire- and drought-tolerant species composition. These conditions are likely to be resilient to changing climate, at least in the short term. Substantial mortality of large trees and continuing loss of basal area, however, are a concern, given further climate warming.
The population irruption and decline of the northern Kaibab deer herd has served as a classic model in wildlife management. The traditional explanation lists the causal factors for deer increase as predator control and protection of does, while the decrease resulted from loss of food supply and habitat caused by overpopulation. Prior grazing by domestic livestock, fire control, and drought also affected the deer population. -Authors