Plant Ecology 163: 187–207, 2002. © 2002 Kluwer Academic Publishers. Printed in the Netherlands.

187

Inter-hemispheric comparison of fire history: The Colorado Front Range, U.S.A., and the Northern Patagonian Andes, Argentina Thomas T. Veblen1 & Thomas Kitzberger2 1 Department

of Geography, University of Colorado, Boulder, CO 80309-0260, USA. (e-mail: [email protected]); 2 Laboratorio Ecotono, Departamento de Ecolog´ıa, Universidad Nacional del Comahue, Argentina (e-mail: [email protected])

Key words: Argentina, Austrocedrus chilensis, climatic variation, Colorado, El Niño-Southern Oscillation, fire, Pinus ponderosa

Abstract Fire history was compared between the Colorado Front Range (U.S.A.) and northern Patagonia (Argentina) by dating fire-scars on 525 Pinus ponderosa and 418 Austrocedrus chilensis, respectively, and determining fire weather on the basis of instrumental and tree-ring proxy records of climatic variation. Years of above average moisture availability preceding fire years, rather than drought alone, is conducive to years of widespread fire in the Colorado Front Range and the northern Patagonian study areas. Above-average precipitation promotes fire by enhancing the growth of herbaceous plants which increases the quantity of fine fuels during the fire season a few years later. The short-term variability in moisture availability that is conducive to widespread burning is strongly related to El Niño Southern Oscillation (ENSO) activity. The warm (El Niño) phase of ENSO is associated with greater moisture availability during the spring in both regions which leads to peaks in fire occurrence several years after El Niño events. The warmer and drier springs associated with la Niña events exacerbate the drying of fuels so that fire years commonly coincide with La Niña events. In both regions, there was a dramatic decline in fire occurrence after the early 1900s due to a decline in intentionally set fires by Native Americans and European settlers, fuel reduction by livestock grazing, and increasingly effective organized fire suppression activities after the 1920s. In both regions there was a marked increase in fire frequency during the mid-and late-19th centuries which coincides with increased ignitions by Native Americans and/or European settlers. However, year-to-year variability in ring widths of Pinus ponderosa and Austrocedrus chilensis also increased from relatively low values in the late 1700s and early 1800s to peaks in the 1850s and 1860s. This implies frequent alternation of years of above and below average moisture availability during the mid-19th century when the frequencies of major fire years rise. The high correlation of tree-growth variability betweem the two regions implies a strong inter-hemispheric variation in climatic variability at a centennial time scale which closely parallels a variety of proxy records of ENSO activity. Based on the relationship of fire and ENSO events documented in the current study, this long-term trend in ENSO activity probably contributed to the mid- and late-19th century increase in fire spread in both regions. These similar trends in fire occurrence have contributed to similar patterns of forest structures, forest health, and current hazard of catastrophic wildfire in the Colorado Front Range and northern Patagonia.

Introduction In this study, we compare fire regimes and their ecological consequences in two broadly similar regions: the eastern foothills of the Front Range of northern Colorado, USA and the eastern foothills of the An-

des in northern Patagonia, Argentina. Despite major differences in the history of taxa, areas of similar regional climate at temperate latitudes in the northern and southern hemisphere tend to be characterized by the same plant life forms arranged into structurally similar assemblages (Box 1981). For example, along

188 the western edges of North and South America there are broadly similar vegetation patterns extending from tropical deserts through Mediterranean-type sclerophyll shrublands and forests to higher latitude rain forests of marine west coast climates (Veblen & Alaback 1996). Comparisons of inter-hemispheric vegetation patterns often reveal insights into the importance of differences in phylogenetic history, finer-scale differences in modern or paleoclimates, or human influences in determining broad-scale vegetation patterns (Mooney 1977; Arroyo et al. 1995). A similar approach based on comparing disturbance regimes and their ecological consequences in regions of similar climate and vegetation physiognomy may yield insights about the dynamics of the respective plant communities and landscapes. For example, temperate rain forests along the west coasts of the Americas and in the southwest Pacific (Australia and New Zealand) have been compared in terms of the relative importance of disturbances such as fine-scale treefalls, large blow downs, geologically-related disturbance (volcanism, earthquakes), and fire (Veblen & Alaback 1996; Brown & Read 1996). Comparison of the consequences of such disturbances to the structure and dynamics of Nothofagus forests across the south Pacific was facilitated by the adoption of the same field sampling and analytical protocols in different regions (e.g., Veblen et al. 1981; Read & Hill 1985). In contrast to the relatively numerous attempts at regional intercomparisons of vegetation patterns and structure of ecoystems on different continents, there are relatively few inter-continental or interhemispheric quantitative comparisons of disturbance regimes. This may be partially due to the more standardized methodologies for describing vegetation physiognomy and ecosystem structure. Reviews of disturbance patterns and their consequences in different regions are often hindered by lack of agreement on standard protocols for collecting, analyzing and interpreting data on disturbance ecology (Barden 1989; Huggard & Arsenault 1999). In the present comparison of fire regimes between Colorado and Argentina, we exploit the fact that similar field and analytical procedures have been applied to the study of fire regimes in both regions (e.g., Kitzberger & Veblen 1997; Kitzberger et al. 1997; Veblen et al. 1999; Veblen et al. 2000). Furthermore, the ecological consequences of changes in fire regimes in both regions have been investigated through similar methods of stand-level age-structure analyses and landscape-level comparisons of historical and modern photographs

(Veblen & Lorenz 1986, 1987, 1991; Mast et al. 1997; Kitzberger & Veblen 1999). The central research questions which have directed this research in both Colorado and northern Patagonia are: (1) How have fire regimes changed over the past ca. 400 years? (2) To what extent can these changes be attributed to variations in climate or human activities? And, (3) how does climatic variation associated with the El Niño-Southern Oscillation (ENSO) affect year-to-year variation as well as longer term variation in fire history? We also briefly review the ecological consequences of changes in fire regimes. In particular, we stress the importance of considering both human and climatic influences on fire regimes in the context of understanding stand- and landscape-scale changes in vegetation structure and forest health. The regions chosen for this comparison are both characterized by conifer-dominated forests ranging from xeric open woodlands to submesic dense stands. In the Colorado study area, the dominant conifer is Pinus ponderosa Laws. var. scopulorum Engelm. (ponderosa pine) and in northern Patagonia it is Austrocedrus chilensis (D. Don) Florin et Boutleje (ciprés). Both study areas are bordered towards the east by semi-arid grasslands (or steppes) of continental interiors and towards the west by higher elevation, more mesic forests in the Rockies and the Andes, respectively. In addition to having similar vegetation physiognomy, the two regions share similar histories of human settlement and both show significant influences of ENSO on the regional climates, as described in the Study Areas. The fire history data presented here are taken from previously published papers (Kitzberger et al. 1997; Kitzberger and Veblen 1997; Veblen et al. 1999; Veblen et al. 2000). The data from these previous studies have been re-compiled into data sets that focus on the P. ponderosa and Austrocedrus vegetation types, and analyzed in relation to new proxy records of climatic variation following common protocols to facilitate comparison of fire history between the two regions.

Study areas Colorado Natural Setting. In Colorado, fire history was sampled in the montane zone (ca. 1,830–2,790 m) of the eastern slope of the northern Colorado Front Range at ca. 40◦ N from near the southeastern border of Rocky

189

Figure 1. Maps showing the locations of the northern Front Range of Colorado, USA, and northern Patagonia, Argentina. Dots are fire history sample sites.

Mountain National Park to southwest of the city of Boulder (Figure 1). Sites sampled for fire history were on City of Boulder Open Space and Mountain Parks lands or on Roosevelt National Forest lands. The bulk of the Front Range in Boulder County is composed of a core of acidic Precambrian intrusive rocks, which in turn have been intruded by acidic Tertiary plutons (Veblen & Lorenz 1991). Towards the east at elevations below ca. 2,200 m, outside of the mineralized belt are a series of narrow sedimentary formations forming cuestas and hogbacks. Soils are highly variable but are usually coarsely textured and shallow. The climate is typical of high elevation, continental regions with strong temperature contrasts between

summer and winter (Greenland et al. 1985). Synopticscale climate is dominated in the winter by westerly flow from the Pacific. When this flow meets the Rocky Mountains, the uplift of the airmass results in precipitation mainly on the western slope. In spring and autumn, occasional eastern upslope conditions develop, pulling moist air up from the Gulf of Mexico, creating heavy precipitation along the eastern slope of the Rockies. In the summer, with the development of the North American monsoon, convection cells create locally intense thunderstorms. Along the eastern slope of the Front Range, there is a relatively gradual increase in precipitation and marked decrease in temperature. In the study area, mean annual temperature ranges

190 from 10.9 ◦ C at an elevation of 1,646 m to 4.7 ◦ C at 2,590 m (data for 1944–1997 and 1945–1993 for Boulder and Allenspark, respectively). Precipitation is less variable with elevation, ranging from 482 mm at 1,646 m to 536 mm at 2,590 m over the same period. In the Colorado Front Range, ENSO events have a significant influence on seasonal temperature and precipitation (Veblen et al. 2000; Donnegan 2000). In association with La Niña events fall and spring temperatures are above average, and are below average during El Niño events (Figure 2). Precipitation tends to be below average during winter through summer in association with La Niña events and the opposite during El Niño events. These patterns imply that La Niña events would be favorable to fuel desiccation during the summer fire season. On the eastern slope of the Colorado Front Range, the structure of Pinus ponderosa-dominated ecosystems varies with elevation, ranging from open woodlands at the ecotone with Plains grasslands (ca. 1,800 m) to dense mixed stands with Pseuodotsuga menziesii (Mirb.) Franco (Douglas fir) and Pinus contorta Dougl. var. latifolia Engelm. (lodgepole pine) at over 2,500 m (Marr 1961). With increasing elevation there is also an increase in the importance of Populus tremuloides Michx. (aspen) and Pinus flexilis James (limber pine). Throughout the elevational range of Pinus ponderosa, north-facing, moister slopes tend to be more densely forested and characterized by greater relative abundances of Pseudotsuga menziesii (Marr 1961). Human settlement and land use Native Americans settled and hunted in the Front Range for at least several thousand years prior to the first permanent settlement by Euro-American settlers in the late 1850s (Buchholtz 1983). There are 19th century reports of Native Americans intentionally setting fires in the Front Range and in northwestern Colorado for driving game animals and occasionally in warfare (Jack 1899; Sudworth 1899). Mineral prospecting in the 1850s yielded a major gold strike in 1858 and subsequent rush of miners and other settlers (Smith 1981). There are numerous reports of early settlers intentionally setting fires during the second half of the 19th century (Tice 1872; Jack 1899; Clements 1910). Livestock numbers are believed to have peaked in the 1890s to early 1900s (Marr 1961), and most of the area has not been grazed since the 1960s. Selective logging of Pinus ponderosa and Pseudotsuga menziesii was widespread from the late 1850s to the

early 1900s, but subsequent logging has been minor (Veblen & Lorenz 1991). Most of the northern Front Range (including present-day Arapaho-Roosevelt National Forest) was included in the Medicine Bow Forest Reserve established in 1905 and subsequently would have been subjected to a fire suppression policy (Veblen & Lorenz 1991). Northern Patagonia Natural setting Most of the sites sampled for fire history are in one of four large Argentine National Parks: Lanín, Nahuel Huapi, Lago Puelo, and Los Alerces National Parks (from ca. 40 to 43◦ S). The study area encompasses the eastern foothills of the Andes and the adjacent Patagonian plains at an elevation of ca. 800 m. Soils are mainly derived from recent volcanic ash that overlie Pleistocene glacial topography. The study area is bounded to the west by the Andean Cordillera which reaches elevations of over 2,000 m and to the east by the Patagonian plains. Seasonal and annual variations in precipitation in northern Patagonia (at ca. 40◦ S) are strongly influenced by changes in the intensity and latitudinal positions of the southeast Pacific anticyclone (Villalba & Veblen 1998). In winter, the subtropical anticyclone is located near 33◦ S off the coast of central Chile, and steers westerly cyclonic storms into northern Patagonia. During the spring and summer months (November-March) the subtropical anticyclone migrates southwards to ca. 40◦ S, where it blocks the westerly flow of moist air masses into northern Patagonia. The Andes have a pronounced rain shadow effect on moist Pacific air masses flowing from west to east. Mean annual precipitation declines from > 3,000 mm in the rain forest zone near the continental divide to ca. 800 mm near the ecotone with the Patagonia steppe (Barros et al. 1983). In the zone of sparse Austrocedrus woodlands, mean annual temperature is 7.9 ◦ C, and mean annual precipitation is 970 mm (data for 1915– 1997 from Bariloche, 41◦ 09 S, 71◦10 W). In northern Patagonia, ENSO events have a significant influence on seasonal temperature and precipitation (Daniels & Veblen 2000) which is more complex and variable than the influence described for Colorado. During La Niña events, spring temperatures tend to be above average but summer-fall temperatures are below average (Figure 2). The opposite is true for El Niño events. During La Niña events, winter-spring

191

Figure 2. Schematic diagram showing documented influences of ENSO events on seasonal temperature and precipitation in the Colorado Front Range and northern Patagonia. The patterns shown are based on statistically significant relationships of measures of the Southern Oscillation (Southern Oscillation Index and tropical Pacific sea surface temperatures) to seasonal temperatures and precipitation in Veblen et al. (2000), Donnegan (2000) or Daniels & Veblen (2000).

precipitation is below average and summer precipitation is above average. The opposite is true for El Niño events. Thus, during the spring-summer fire season, fuel desiccation could result from below average spring precipitation (associated with La Niña) or from warmer than average summers (associated with El Niño). The strong west-to-east decline in precipitation east of the Andes is reflected in a steep vegetation gradient from temperate rain forest to the Patagonian steppe (Veblen et al. 1996). Towards the west, rain forests are dominated by 40 to 45-m tall evergreen Nothofagus dombeyi Mirb. (Oerst.). Bordering the rain forest, towards the east, Austrocedrus chilensis and N. dombeyi form extensive stands. Pure, dense stands of Austrocedrus occupy more xeric sites. At mean annual precipitation of 800 to 1,200 mm, Austrocedrus forms sparse woodlands near the ecotone with the Patagonian steppe (Barros et al. 1983). The steep precipitation and vegetation gradient from the steppe to

the Andean rain forest is an important contrast with the gradual increase in precipitation along the elevational gradient from the Plains to the Rocky Mountain subalpine zone. However, the present analysis includes only areas of climatically comparable xeric woodland and forests dominated by Pinus ponderosa in Colorado or by Austrocedrus chilensis in Patagonia. Human settlement and land use Prior to permanent white settlement in the 1890s, Native American hunters used fire to hunt guanacos (Lama guanicoe) and other game mainly in the xeric, open Austrocedrus woodlands and adjacent steppe (Cox 1863; Fonck 1900). In the 1850s, the occupation of northern Patagonia by Native American hunters increased due to European settlement of adjacent southern Chile and the emigration of many Native Americans across the Andes into Argentina (Cox 1863). Although it is known that Native Americans ignited fire, it is not known how abundant those ignitions were relative to lightning-caused fires. Be-

192 tween the late 1880s and the 1920s, European settlers burned extensive areas of Andean forest in this region in order to clear forests and establish livestock raising (Willis 1914; Rothkugel 1916). Livestock populations grew rapidly after about 1890 and large populations had free access to most of the study area for most of the first half of the 20th century (Veblen & Lorenz 1988). During the second half of the century livestock were increasingly restricted in numbers and location. Logging of Austrocedrus was extensive through the 1940s and subsequently became restricted to limited areas. The first national park in the region was created in 1922, but infrastructure for controlling fires was relatively ineffective until at least the 1930s.

Methods Fire history methods Fire-scarred trees were sampled at 41 Pinus ponderosa sites in Colorado and at 18 Austrocedrus chilensis sites in northern Patagonia (Figure 1). Sample sites ranged in elevation from 1,884 to 2,670 m in Colorado and from 700 to 1,100 m in northern Patagonia. The Colorado sites ranged from relatively open woodlands purely of P. ponderosa at low elevations to dense stands mixed with mainly Pseudotsuga menziesii at higher elevations. Likewise, in northern Patagonia sample sites ranged from xeric woodlands of Austrocedrus at low elevations and easterly locations to dense stands of Austrocedrus mixed with Nothofagus dombeyi towards the west. Individual sample sites were of similar size (ca. 50 to 200 ha). The goal was to collect at least 10 fire scars per site without exceeding a search area of 200 ha. However, the number of trees sampled per site ranged from 6 to 49 depending on the availability of fire-scarred trees. All sample areas were searched for relatively old fire-scarred trees which are scarce in both regions due to mortality caused by logging, fires and insect pests. Partial cross sections were cut from live and dead trees, and fire-scar dates were determined by visual and quantitative cross-dating against marker rings from master tree-ring chronologies at nearby sites. The numbers of cross sections that yielded annually precise fire dates were 525 and 418 for Colorado and northern Patagonia, respectively. According to convention, the calendar dates of annual rings in the southern hemisphere are assigned to the year in which ring formation begins. Thus, a fire that occurs in

mid-summer (February) would be assigned to the preceding calendar year because tree rings begin to form in October to December. Further details on the fire history sample sites and methods are given in Kitzberger et al. (1997), Kitzberger & Veblen (1997), and Veblen et al. (1999, 2000). The computer program FHX2 (Grissino-Mayer 1995), an integrated software package for analysis of fire history information from tree rings, was used to describe and analyze trends in fire-scar dates. Since trees that have been initially fire scarred are more likely to record subsequent fires due to the loss of some protective bark, trees are considered recorder trees only from the date of their initial scarring until the date of their death or date of field sampling. Years of fire occurrence were determined according to criteria indicating a gradient from years of minor to widespread fire occurrence: (1) years in which any fires were dated; (2) years in which a minimum of 2 recorder trees was scarred in at least one site; and (3) years in which scars were recorded on higher percentages of trees (e.g., ≥ 10% and 25% at a site with a minimum of 20 recorder trees). Temporal trends in fire occurrence were examined by plotting annual time series of the percentages of sites that recorded fires from 1600 to the 1990s. Due to the mortality and decay of old trees, evidence of old fires gradually disappears. Consequently, we used percentages of recorder sites (i.e. sites containing a dated fire-scar that pre-dates the year plotted) that record a fire date instead of absolute numbers of sites. This procedure, as well as the beginning of the plots in 1600 when each region contained at least 8 recorder sites, reduces the bias due to disappearance of older evidence of fire. Relationships of fire to climatic variation Mean climatic parameters were compared for fire years and non-fire years based on both instrumental climatic records and a climatically-sensitive tree-ring record. Superposed epoch analysis (Baisan & Swetnam 1990; Grissino-Mayer 1995) was used to test the null hypothesis that there is no relationship between occurrence of fire years and climatic conditions in the years preceding and during fire years. Mean values of climatic parameters from instrumental records or tree-ring proxy records (described below) were calculated for 5 to 9-year windows including the year of the fire event. Mean values of climatic parameters preceding, during and following fire years were compared

193 to variation in the complete record by performing Monte Carlo simulations that randomly pick years, calculate expected means, and provide 95% bootstrap confidence intervals (Mooney & Duval 1993; Grissino-Mayer 1995). In each case, the number of randomly selected years equals the number of actual fire years. Results are described as percentage departures from the mean values determined by the random selection of non-fire years. The instrumental climate records used were the longest and most complete records available in each region. The Denver record begins in 1872, and the Bariloche record begins in 1906 for precipitation and 1915 for temperature. Temperature and precipitation were seasonalized into two and three month seasons for comparison of mean seasonal climatic conditions during and prior to years of any fire occurrence. Results are presented only for the spring and summer because they had the strongest relationships with fire occurrence. To permit analyses of fire-climate relationships over a multi-century period, regional tree-ring records were used as indicators of year-to-year variations in temperature and precipitation. The regional tree-ring chronology in Colorado consisted of 199 tree cores, collected from 10 to 17 trees at each of eight sites of Pinus ponderosa forests or woodlands located at 39◦41 to 40◦25 N in the northern Front Range. For Patagonia, the regional record consisted of 585 tree cores collected from 8 to 30 trees at 25 sites of Austrocedrus woodlands at 37◦ 04 to 43◦ 11 S. Details of methods used for creating the Pinus ponderosa and Austrocedrus regional chronologies and calibrating them against instrumental climatic records are given in Villalba & Veblen (1997b) and Veblen et al. (2000), respectively. Both of these regional chronologies are robust indicators of spring and summer moisture availability. Thus, narrow rings indicate below average precipitation and/or above average temperatures during spring-summer. The first differences (i.e., difference between tree-ring index in year t and year t−1) were computed as a measure of the year-to-year variability in tree growth (and by inference moisture availability). To assess long-term changes in year-to-year climatic variability, moving fifty year averages of the first differences were computed and their departures (standard deviations) were plotted. Two types of records of ENSO activity were used to examine possible influences of ENSO fire occurrence. The first record is a 1706–1977 reconstruction of the December-February Southern Oscillation Index

(SOI) based on tree-ring chronologies from ENSOsensitive regions of subtropical North America and Indonesia (Stahle et al. 1998). This reconstruction explains 53% of the variance in instrumental DecemberFebruary SOI based on differences in standardized sea-level pressure between Tahiti and Darwin, Australia (Ropelewski & Jones 1987). High values of the index are indicative of the positive phase of the Southern Oscillation (La Niña). The second record is Quinn’s (1992) documentary compilation of El Niño years from 1525 to 1987. From the Quinn record, all years of ‘strong’ and ‘very strong’ El Niño events were used as event years in superposed epoch analysis of variations in the percentages of sites recording fire scars. The Quinn (1992) record is qualitative and is not as annually precise as the tree-ring reconstruction of SOI (Stahle et al. 1998) because El Niño events may start in different months and may span more than a single year. However, the Quinn (1992) documentary record of El Niño years is based on entirely independent data from the tree-ring reconstruction of SOI. Results and interpretation Fire seasonality The Colorado and Patagonia study areas differ in the seasonal distribution of precipitation and degree of continentality (Figure 3). Precipitation in Bariloche is high during the late fall and winter (June–September) and relatively low during the spring and summer. In contrast to this mediterranean-type distribution of precipitation, in Colorado precipitation reaches a peak in the spring months of April–May and is at its lowest during the winter months of December through February. Although the annual precipitation in northern Patagonia is approximately twice as great as in Colorado, moisture availability during the six month fire season is similar in both areas. Martonne’s (1926) aridity index averages 2.2 and 2.5 for the six-month fire seasons in Colorado and northern Patagonia, respectively. Somewhat higher spring precipitation in Colorado is balanced by the higher summer temperatures of this region. During the two driest summer months, moisture availability on average is virtually the same in both areas (Martonne’s aridity index = 1.6) The similarity of seasonal moisture availability is reflected by similar seasonal patterns of fire occurrence. In Colorado, over the period 1915 to 1992,

Figure 3. Seasonal distribution of climatic parameters and fire occurrence in the northern Front Range of Colorado (a and c) and in northern Patagonia (b and d). Fire data are from the Argentine National Parks of Lan´ın, Nahuel Huapi, Lago Puelo and Los Alerces for 1940–1988 (n = 470) and from Rocky Mountain National Park for 1915–1992 (n = 136). Martonne’s monthly aridity index = Pm∗ 1.2 / (Tm + 10) where Pm is monthly precipitation in mm and Tm is mean monthly temperature in ◦ C. Note that the axes for the aridity indices are inverted and scaled differently.

194

195 97.8% of the fires occurred during May–October and 58.6% occurred in the two peak summer months of July and August (Figure 3). In northern Patagonia, over the period 1940–1988, 95.3% of the fires occurred during October–March and 58.7% occurred in the two peak summer months of January and February. Influences of inter-annual climatic variation on fire occurrence Relationships of fire to modern instrumental records of climatic variation There are both similarities and differences in the relationships of years of fire occurrence to variations in seasonal precipitation and temperature in Colorado and Patagonia during the periods of the instrumental climate records (i.e., after 1872 in Colorado and after 1906 in Patagonia). Fire years in Colorado coincide with years of below average spring and summer precipitation but are not dependent on above average temperature during the summer of the fire season (year t in Figure 4a). Warm springs may contribute to fire occurrence in Colorado but the relationship is not statistically significant (Figure 4a). In contrast, in Patagonia fire years do not show a significant relationship to precipitation during the spring and summer of year t (Figure 4b). They are, however, strongly associated with warm summers (Figure 4b). Thus, the desiccation of fuels necessary for significant fire spread is clearly associated with low spring-summer precipitation in Colorado whereas in Patagonia, variance in summer temperature plays a greater role. Lightning is frequent every summer in Colorado and is unlikely to a be a limitation on fire occurrence. In Arapaho-Roosevelt National Forest which surrounds the study area in Colorado, lightning accounted for 33.7% of the number of fires from 1909 to 1991 (n = 3, 352 fires) and 33.3% of the area burned from 1924 to 1998 (total area burned = 14,355 ha; data from USDA Forest Service). Although most modern fires in Patagonia are set by humans, lightning is an important source of ignition. Lightning ignited 8% of all fires (n = 722 fires) and accounted for 16.4% of the area burned (total area burned = 118,560 ha) in the four national parks from 1938 to 1996 (data from Administración de Parques Nacionales). In contrast to Colorado, in Patagonia, lightning is rare or absent in some years, and the association of lightning with years of higher summer temperatures contributes to the association of fire years with warmer temperatures (Kitzberger et al. 1997).

In both Colorado and Patagonia, fire years follow by one to several years weather conditions which would be favorable to fuel production. In Colorado, years of significant fire occurrence follow above average spring precipitation by four years and below average summer temperatures by one year (Figure 4a). In Patagonia, years of significant fire occurrence follow above average spring precipitation by three and four years (Figure 4b). In both regions, however, the numbers of fire years that overlap the instrumental climate records prior to fire suppression are relatively low (Figure 4). Consequently, it is important to consider the longer term pattern of fire occurrence and climatic variation that can be inferred from tree-ring proxy records of moisture availability. Relationships of fire years to long-term proxy records of climatic variation Tree-ring records of spring-summer moisture availability indicate a pattern of fire-promoting weather similar to the pattern indicated by the instrumental record of precipitation and temperature. Over the periods from 1600 to the approximate date of effective fire suppression in both regions (1920 in Colorado and 1929 in Patagonia), tree-ring proxy records show that years of fire occurrence were below average in moisture availability (Figure 5). In both regions, years of fire occurrence follow by 2 or 3 year periods of above average moisture availability. Years of fire occurrence in both Colorado and Patagonia are significantly associated with the variations in the tree-ring reconstruction of December– February SOI (Figure 6). In Colorado, fire years coincide with high SOI indicating that fires co-occur with La Niña conditions. This is consistent with the warmer springs and drier spring-summers associated with La Niña events in Colorado (Figure 2). In Patagonia, although the year of the fire event also tends to have above average SOI, the highest SOI occurs during the year prior to the fire event (Figure 6). This implies that fire event years tend to immediately follow or occur during the late stages of La Niña events. The documentary record of El Niño also indicates a marked influence of ENSO activity on fire activity in both regions (Figure 7). For Colorado, the percentage of sites recording fire rises from below average during El Niño events to a sharp peak 3 to 4 years after El Niño events (Figure 7a). In Patagonia, fire occurrence also peaks 3 to 4 years following El Niño (Figure 7b). In Patagonia, however, the year of the El Niño event

Figure 4. Spring and summer temperature and precipitation departures in relation to fire years (t) for Colorado (a) and Patagonia (b). Fire years are years in which 2 or more trees recorded fire scars during the periods of overlap with the instrumental meteorological records of Denver, Colorado (1872–1995; n = 29) and Bariloche, Argentina (1906–1996 and 1915–1996 for precipitation and temperature, respectively; n = 35 and 27, respectively). Departures are percentage differences from means derived from a random selection of the same number of years as fire years. For Colorado, Sp (spring) is May–June and Su (summer) is July–August; for Patagonia, Sp is October–November and Su is January–March. Dots indicate statistical significance determined from bootstrap 95% confidence intervals based on 1,000 Monte Carlo simulations of the same number of years as fire years (Mooney & Duval 1993).

196

197

Figure 5. (a) Departure (%) from the mean Pinus ponderosa composite tree-ring index for lag years −4 through +2 for fire years (year 0) with ≥ 2 fire-scarred trees in at least one site for 1600–1920 in Colorado; and (b) departure (%) from the mean Austrocedrus chilensis composite tree-ring index for lag years −4 through +2 for fire years (year 0) with ≥ 2 fire-scarred trees in at least one site for 1600–1929 in Patagonia. Positive departures in both (a) and (b) indicate above average moisture availability. Dots indicate statistical significance determined from bootstrap 95% confidence intervals based on 1,000 Monte Carlo simulations of the same number of years as fire years (Mooney & Duval 1993). The number of fire years is given by n.

and the previous year (years 0 and −1 in Figure 7b) also show peaks of fire occurrence. This strong contrast with the pattern for Colorado is consistent with warmer than average summers in Patagonia associated with El Niño events (Figure 2). Decadal and centennial trends in fire history At decadal to centennial time scales, the trends in percentages of sample sites recording fire scars are similar for Colorado and Patagonia (Figure 8). In both

Figure 6. Departure of the tree-ring proxy record of winter (December–February) Southern Oscillation Index (SOI) for lag years from −4 through +2 for years with ≥ 2 fire-scarred trees in at least one site for 1706–1977 in Colorado (a) and in Patagonia (b). Positive departures indicate the positive (Niña) phase of the Southern Oscillation. The tree-ring proxy of winter SOI is from Stahle et al. (1998). Dots indicate statistical significance determined from bootstrap 95% confidence intervals based on 1,000 Monte Carlo simulations of the same number of years as fire years (Mooney & Duval 1993). The number of fire years is given by n.

regions, the frequency of years with any fire scars begins to decline after ca. 1920 and is sharply lower after ca. 1950. In each region, the beginning of the decline in fire occurrence follows the creation of national forests and national parks by about two decades. A second pattern common to both regions, is an increase in the frequency of years with fire scars beginning in the mid-19th century, in the 1840s in Colorado and in the 1850s in Patagonia (Figure 8). This is a pattern of both increased frequency of fire years and increased extent of fire as indicated by the high per-

198 year-to-year variability in tree-ring widths in both regions (Figure 9c–d). This implies that spring-summer moisture variability, the main determinant of radial growth of both Pinus ponderosa and Austrocedrus, as well as the main determinant of fine fuel production and desiccation, increased during the 19th century. Trends in tree-growth are highly similar, especially after 1700 when both records contain large numbers of samples in Colorado and in Patagonia (r = 0.73). Analogously, the numbers of strong and very strong ENSO events, as reconstructed from historical records (Quinn 1992), follow a similar pattern of increase during the 19th century (Figure 9e).

Discussion Influences of inter-annual climatic variability on fire regimes

Figure 7. Departures (%) from mean numbers of sites with fire-scarred trees for lag years from −3 through +6 from years listed by Quinn (1992) as strong or very strong El Niño events for the period 1600 to 1996 for Colorado (a) and Patagonia (b). Dots indicate statistical significance determined from bootstrap 95% confidence intervals based on 1,000 Monte Carlo simulations of the same number of years as fire years (Mooney & Duval 1993). The number of fire years is given by n.

centages of sites recording fire in some individual years. In Colorado, there is a sharp increase in the number of years with > 10% sites recording fire beginning in 1859. In Patagonia the pattern is similar. Although the 1750 to 1850 period records only a single year with more than 30% of the sites recording fire, 1850 to 1910 period contains 6 such years (Figure 8). In addition to these coincidences of trends in fire occurrence and land use, fire trends also coincide with centennial-scale variations in tree-ring proxy records of climatic variation and in historical reconstructions of ENSO events (Figure 9). In both Colorado and Patagonia, at a time-scale of 50 years, the mid- and late-19th century is a time of frequent occurrence of widespread fires (Figures 9a–b). The increase in fire occurrence approximately parallels an increase in

Both instrumental and tree-ring proxy records of climate indicate that years of above average moisture availability preceding fire years rather than drought alone is conducive to years of widespread fire in the Colorado Front Range and northern Patagonian study areas. For example, these records indicate that fire occurrence, especially years of more widespread fire, tend to increase three to four years following aboveaverage precipitation during the spring (Figure 4). In both regions, plant growth is highly sensitivity to variations in spring moisture availability (Graybill 1992; Villalba & Veblen 1997b; Veblen et al. 2000). Aboveaverage precipitation promotes fire by enhancing the growth of herbaceous plants which increases the quantity of fine fuels during the fire season a few years later. In Colorado, cooler temperatures the year prior to fire occurrence may also enhance fuel accumulation (Figure 4a). Similar associations of fire, with fuelenhancing wet periods one to several years prior to burn events, have been reported for other ecosystems where seasonal drought appears to restrict fire through limitation of fuel quantity (Swetnam & Betancourt 1998; Grau & Veblen 2000). The Colorado Front Range and northern Patagonia differ in the relative importance of precipitation and temperature variations in creating sufficiently dry conditions for fire occurrence. In Colorado, years of fire occurrence coincide with below average spring and summer precipitation but do not appear dependent on above average summer temperatures (Figure 4a). Given the higher average summer temperatures of the

199

Figure 8. Percentages of sites with fire-scar dates per year in Colorado (a) and Patagonia (b). Sample depth (stepped horizontal lines) gives the number of sites with recorder trees alive in each year. The dashed line indicates one standard deviation above the mean annual percentage of sites with scarred trees. Numbered triangles indicate approximate dates of major changes in land use in each region: (1) beginning of the mineral rush in Colorado (1858–59); (2) establishment of the national forest reserves in Colorado (1905); (3) increased settlement by Native Americans in Patagonia (1850s); (4) European settlement and forest clearing (1880s to ca. 1910; and (5) creation of the first national park in Patagonia (1922).

more continental climate of Colorado, favorable fire weather is less dependent on anomalous summer temperatures. In contrast, in Patagonia, the instrumental record indicates that fuel desiccation during fire years is more strongly related to warm summer temperatures than to summer precipitation (Figue 4b). In more mesic forests (west of the xeric Austrocedrus woodlands), however, fire is dependent on below average spring precipitation (Kitzberger et al. 1997; Veblen et al. 1999). For both the Colorado Front Range and northern Patagonia, the short-term variability in moisture

availability that is conducive to widespread burning is strongly related to ENSO activity. Instrumental records of climatic variation show that the warm (El Niño) phase of ENSO is associated with greater moisture availability during the spring in both regions (Figure 2). Thus, enhanced production of fine fuels associated with El Niño events appears to explain the peak in fire occurrence several years following El Niño events (Figure 7). The similar dependence of plant growth on spring moisture availability and its similar relationship to ENSO events results in peaks in fire

200

Figure 9. Comparison of centennial scale variations in years of major fires in Colorado (a) and Patagonia (b), year-to-year variability in tree growth in Colorado (c) and Patagonia (d) and El Niño events (e). (a) and (b) are the moving 50-year sums of years in which > 25% of Pinus ponderosa and > 10% of Austrocedrus chilensis recorder trees record fire dates in the same year, respectively. (c) and (d) are departures (in standard deviations) of the moving 50-year annual averages of first differences of standardized tree-ring indices of regional tree-ring chronologies of Pinus ponderosa and Austrocedrus chilensis, respectively. (e) is the moving 50-year average number of El Niño years reconstructed from historical sources. Dashed lines are the means for each parameters. The data for (a) and (c) are from Veblen et al. (2000), for (b) from Kitzberger & Veblen (1997) for (d) from Villalba & Veblen (1997b) and for (e) from Quinn (1992).

201 occurrence a few years after El Niño events in both regions. The desiccating effects of an ENSO event can differ between the two regions. The Colorado Front Range follows a simple pattern in which the warmer and drier springs associated with la Niña events exacerbate the drying of fuels that consistently occurs under the high summer moisture deficits of the more continental climate (Figures 2 and 6a). In northern Patagonia, the pattern is more complex and variable. Monthly SOI, over the period of 1882 to 1989, indicates that major fire years tend to coincide with the late stages of La Niña events (Kitzberger & Veblen 1997). This is consistent with the peak in reconstructed SOI during the year prior to fire events (Figure 6b) and with reduced spring moisture availability during La Niña events. However, there is also a peak in fire occurrence which coincides with El Niño years (Figure 7b). This apparent contradiction is explained by the observation that depending on the timing of ENSO events, increased fire in Patagonia is associated with either El Niño or La Niña events. Although the most common pattern is for years of major fire activity to occur during the late stages of La Niña events (Kitzberger & Veblen 1997), major fire years also can coincide with El Niño events (Veblen et al. 1999). In some cases, this may be the same pattern (i.e., transition from La Niña to El Niño events), but in other cases a major fire year can coincide with an El Niño event in the absence of a preceding strong La Niña event. Among the 10 years of greatest fire occurrence between 1740 to 1995, six years coincide with moderate to very strong El Niño events. The timing of ENSO within the annual cycle is key to determining its influence on fire occurrence in northern Patagonia. El Niño events most frequently begin to develop in March to May (Kiladis & Diaz 1989), which for northern Patagonia tends to increase winter-spring (June–November) precipitation during the same calendar year. However, El Niño events that begin to develop after the winter rainy season (e.g., after October), do not result in increased winter-spring precipitation until after the summer dry season. Thus, the warm temperatures associated with late developing El Niño events more effectively desiccate fuels and can promote widespread fire. A transition from a La Niña event (or ‘Niña-like conditions’ of above average SOI) to El Niño conditions promotes fire occurrence because warm summers tend to follow winter-springs of normal or below average precipitation. Thus, there are two ENSO-related weather conditions associated

with years of extreme burning: (1) warmer summers associated with El Niño events that develop after the preceding winter rainy season; and (2) reduced winterspring precipitation during La Niña events preceding the summer fire season. Despite the association of 6 of the 10 most extreme fire years from 1520 to 1929 with El Niño events (that probably developed late in the calendar year), most major fire years between 1520 and 1929 (n = 88) are the year prior to the beginning of El Niño events (Veblen et al. 1999). This is consistent with the tendency of ENSO to switch from one extreme to the other in consecutive years (Diaz & Kiladis 1992), so that many of these major fire years would have followed dry La Niña winter-springs. Some of the others would have been associated with early stages of late developing El Niño events. Long term trends in the fire regimes For both regions, the most dramatic temporal change in fire occurrence is the abrupt decline in fire frequencies in the early 1900s. This is consistent with fire history studies conducted in nearby Pinus ponderosa ecosystems in the Colorado Front Range (Rowdabaugh 1978; Laven et al. 1980; Skinner & Laven 1983; Goldblum & Veblen 1992; Brown et al. 1999; Donnegan 2000). Several factors contributed to this decline in burning during the 20th century: (1) In both regions the Native American hunters intentionally set fire to the vegetation to drive their prey (Jack 1899; Cox 1863; Fonck 1900). In the absence of pre-20th century data on extent of fire spread in relation to sources of ignition (human versus lightning), it is impossible to quantitatively assess the degree to which Native Americans increased the extent of burning over that which would have resulted from lightning alone. However, the demise of these hunting populations resulted in a decline in numbers of ignitions. (2) In both regions, early European settlers intentionally set fires for forest clearance and ranching activities (Jack 1899; Willis 1914). (3) Livestock raising probably reduced availability of grass fuels which in some of the drier phases of these woodland ecosystems may have contributed to a decline in fire spread. (4) Finally, organized fire suppression activities became increasingly more effective with the development of better infrastructure for fire detection and fire control after the 1920s. The increased fire frequency in northern Patagonia during the mid- and late-19th centuries in both regions coincides with changes in land use as well as changes

202 in ENSO activity. Fire frequencies began to increase at most Austrocedrus woodland sites after ca. 1850 and peaked in the late 19th century (Figure 7). The mid19th century increase in all fires is coincident with increased use of the Austrocedrus habitat by Native American hunters as a result of immigration from the Chilean side of the Andes, stimulated by the European colonization of southern Chile (Cox 1863). Spatial variation in the degree of increase of small fires (as opposed to years of widespread burning), according to human access, suggests that the mid-19th century increase in burning is at least partially attributable to an increase in human-set fires (Veblen et al. 1999). High fire frequencies in the 1890s to early 1900s coincide with the period of extensive forest burning by European settlers in attempts to create cattle pasture (Willis 1914; Rothkugel 1916). The effects of European settlement are also reflected by increased fire scars in rain forests along Andean travel routes in the early 1900s (Kitzberger et al. 1997; Veblen et al. 1999). The similar mid- and late-19th century increase in fire occurrence for the Colorado Front Range (Figure 7) is also recorded in other fire history studies conducted in Pinus ponderosa forests in the Front Range (Rowdabaugh 1978; Laven et al. 1980; Skinner & Laven 1983; Goldblum & Veblen 1992; Donnegan 2000). In Colorado, the sharp increase in years with ≥ 2 trees scarred in 1859–1860 (Figure 7), immediately follows the gold discoveries of 1858 and 1859 that triggered a major gold rush in the Front Range (Smith 1981; Buchholtz 1983). Furthermore, there are numerous historical observations of widespread fires in the northern Front Range during the mid1800s period of exploration and early settlement that were allegedly set by Euro-American settlers (Tice 1872; Jack 1899; Clements 1910). However, the trend towards increased burning starts in the 1840s (Figure 7a), at least a decade before the founding of mining towns in 1859–1860. The 1840s was a period of active prospecting and exploration by Euro-Americans (Hafen & Hafen 1956; Smith 1981; Buchholtz 1983). Although increased frequency of human-set fires appears to have contributed to the increase in fire occurrence after the mid-19th century in both regions, climatic conditions favorable to fire spread also appear to have characterized this time period. Year-to-year variability in ring widths of Pinus ponderosa and Austrocedrus chilensis increase from relatively low values in the late 1700s and early 1800s to peaks in the 1850s and 1860s (Figure 9c–d). This implies frequent alternation of years of above and below average moisture

availability during the mid-19th century when the frequencies of major fire years rose. The high correlation of tree-growth variability in the two regions implies a strong inter-hemispheric variation in climatic variability at a centennial time scale. The historical record of El Niño events (Figure 9e), as well as proxy records, indicate that in comparison with the late 18th century and early 19th century, after ca. 1840, there was an increase in frequency and amplitude of El Niño events (Michaelson & Thompson 1992; Dunbar et al. 1994; Villalba 1994). Based on the relationship of fire and ENSO events documented in the current study, this long-term trend in ENSO activity probably contributed to the mid-19th century increase in fire spread in both regions. The strong influence of high frequency climatic variability on fire regimes in this study supports a similar conclusion about climatic variability and fire frequency in a review of Holocene-length pollen and charcoal sediment studies in Patagonia (Markgraf & Anderson 1994). Years of widespread fire are relatively infrequent from the late 1700s into the early 1800s in both the Colorado Front Range and northern Patagonia (Figure 9a–b), and many fire history chronologies from individual sites in both regions record relatively long intervals between significant fire events during this period (Kitzberger & Veblen 1997; Veblen et al. 1999, 2000). The lower year-to-year variability in tree growth during the same period is consistent with a reduced frequency or amplitude of ENSO events which affect seasonal temperatures and precipitation in both regions. Based on the relationships of ENSO events to seasonal precipitation and temperature patterns in both regions, decades of fewer or less extreme ENSO events would be expected to create fuel conditions less favorable to fire spread. Similar trends of fire occurrence and tree growth variability in relation to centennial scale variations in ENSO activity have been reported for the Southwestern U.S. (Swetnam & Betancourt 1998; Grissino-Mayer & Swetnam 2000). Furthermore, trends in fire history in relation to ENSO variability in the Southwestern U.S. and northern Patagonia are highly similar (Kitzberger et al. 2001). Although ENSO has been an important influence on the climate and apparently the fire history of both the Colorado Front Range and northern Patagonia, the strength of that influence has clearly varied at decadal to centennial time scales (Villalba & Veblen 1998; Daniels & Veblen 2000). ENSO variability during some time periods may result in similar pat-

203 terns of tree growth and fire occurrence in northern Patagonia and parts of western North America, but sources of climatic variability associated with high latitude circulation patterns also influence the trends described here. For example, major fire-event years in northern Patagonia are also associated with belowaverage atmospheric pressure at 50–60◦ S in the South American-Antarctic Peninsula sector of the Southern Ocean (Veblen et al. 1999). Although these high latitude circulation anomalies may also be linked to variations in tropical atmosphere-ocean interactions, such linkages are not yet clear (Villalba et al. 1997). Ecological consequences of altered fire regimes The similarities in fire regimes over the past ca. 200 years are reflected by similar changes in stand- and landscape-scale vegetation patterns in the Pinus ponderosa habitat of Colorado and the Austrocedrus habitat of northern Patagonia. The most dramatic changes in both landscapes are associated with reduced fire frequency during the 20th century. In Colorado, comparison of historical and modern landscape photographs as well as analyses of tree population age structures document the spread of Pinus ponderosa and Pseudotsuga menziesii into some former grasslands (Veblen & Lorenz 1991; Mast et al. 1997, 1998). This major change in montane forests is mainly attributed to elimination of the frequent surface fires that formerly prevented seedling survival at most sites. However, climatic variability and changes in livestock and other herbivore populations also may have influenced the expansion of Pinus ponderosa woodlands (Mast et al. 1998). Increased density of Pinus ponderosa stands during the period of fire exclusion has changed the susceptibility of these forests to stand-replacing fires, pathogen infestation, and perhaps insect outbreak. Qualitatively, it is obvious that very sparse Pinus ponderosa woodlands did not support crown fires due to lack of woody fuel continuity. Quantitatively, however, the increase in area newly capable of supporting crown fires is difficult to estimate because even prior to any significant effects of fire exclusion, some stand-replacing fires occurred in denser Pinus ponderosa forests (Veblen & Lorenz 1986; Shinneman & Baker 1997). At some sites, initially open stands of Pinus ponderosa now have understories of suppressed Pseudotsuga menziesii that are susceptible to insect-caused mortality which may further increase the hazard of stand-replacing fires. Dwarf mistletoe

(Arceuthobium spp.) populations accumulate in both Pinus ponderosa and Pinus contorta stands where fires have been excluded; the weakening effects of these hemi-parasites may increase stand-susceptibility to other pests and fire (Zimmerman & Laven 1984; Kipfmueller & Baker 1998). Although both of these changes in forest health are likely to have occurred in some stands, the lack of precise information on pre-20th century forest health makes it impossible to quantitatively assess the magnitude of these apparent changes in forest health at landscape or regional scales. Analogously, in northern Patagonia, the 20th century period of reduced fire frequency coincides with expansion of Austrocedrus into former grasslands, generally increased tree densities, and conversion from shrubland to forest (Veblen & Lorenz 1988; Kitzberger & Veblen 1999). Comparison of historical and modern photographs, as well as age structure data, indicate that abundant establishment of Austrocedrus began at the ecotone in the late 1800s (Veblen & Lorenz 1988). Reduction in fire occurrence after the early 1900s allowed much greater survival of juveniles of this fire-sensitive species. The rate and timing of Austrocedrus establishment have been influenced by inter-annual and decadal scale climatic variation (Villalba & Veblen 1997a). Indeed, since the increase in frequency of strong El Niño events after 1976, there has been a marked decline in survival of Austrocedrus seedlings in xeric habitats associated with summer heat and moisture stress (Villalba & Veblen 1997a; Kitzberger et al. 2000). The process of tree invasion of Patagonian grasslands during the 20th century period of reduced fire frequency was also modified by livestock impacts (Veblen et al. 1992; Relva & Veblen 1998). The overall increase in tree density in Austrocedrus woodlands and invasion of trees into the steppe have created more contiguous woody fuels (Kitzberger & Veblen 1999) so that sites previously supporting only surface fires are now susceptible to stand-replacing fires. As in Colorado, decreased fire frequency in Patagonia may have increased the incidence of mistletoe infestations. However, in this case the infestation by mistletoes (Misodendrum spp.) affects mainly Nothofagus antarctica (Forst.) Oerst. which grows in woodlands and shrublands either mixed with or adjacent to Austrocedrus stands. Fire exclusion probably has increased Misodenrum loads on N. antarctica. This hemi-parasite increases branch mortality and facilitates fungal and boring insect attacks (N. Tercero &

204 T. Kitzberger unpublished data), chronically infecting stems of Nothofagus, and possibly enhancing post-fire self-thinning. Thus, under fire exclusion, increased infection by Misodendrum may in turn have increased quantities of dead fuels, thereby increasing the hazard of more intense crown fires. Fire exclusion also has decreased the abundance of vigorous post-fire resprouts of N. antarctica which typically is a short-lived tree. Thus, under fire exclusion, both increased infection by Misodendrum as well as reduced vegetative reproduction, may contribute to the widespread stand dieback of N. antarctica (Veblen & Lorenz 1988; Veblen et al. 1996). In both the Colorado Front Range and northern Patagonia, increased hazard of stand-replacing fires in formerly more open conifer woodlands has coincided with dramatic increases in human residential use of these habitats during the late 20th century. Thus, in both regions, fire exclusion has had the long-term effect of severely increasing fire hazards to human life and property. In northern Patagonia, this hazard has been further exacerbated by the planting of introduced conifers (including Pinus ponderosa) in large areas of former steppe. In the context of global warming and the tendency for El Niño events to be more frequent during warmer decades (Tsoni et al. 1998), the climatic component of this fire hazard is likely to increase. The second pattern of altered fire regimes which has had major ecological consequences in the Colorado Front Range and in northern Patagonia is the sharp increase in fire frequency during the latter half of the 19th century. Whether caused primarily by climatic variation or increased ignitions by humans, the net result has been a concentration of widespread fires in a relatively short period of a few decades. The vegetation response in both regions has been the establishment of extensive, even-aged forests dating from the late 1800s to early 1900s period of widespread forest clearing and burning. This response is characteristic of the more mesic habitats of Pinus ponderosa and Austrocedrus. In Colorado, the upper montane zone of Pinus ponderosa mixed with Pseudotsuga menziesii and other conifers consists mostly of even-aged stands dating from the ca. 1880 to 1910 (Clements 1910; Veblen & Lorenz 1986; Hadley & Veblen 1993; Mast et al. 1998). These even-aged stands tend to have understories of suppressed Pseudotsuga menziesii saplings which are highly susceptible to defoliation by western spruce budworm (Choristoneura occidentalis). In-

creased 19th century burning in the upper montane zone created extensive areas of post-fire even-aged stands that more or less synchronously become susceptible to budworm outbreaks (Hadley & Veblen 1993). Tree-ring reconstructions of the history of budworm outbreaks indicate that late 20th-century outbreaks are synchronous over large areas which may be a reflection of the synchronization of stand structural stages by widespread burning in the late 19th century (Swetnam & Lynch 1989; Hadley & Veblen 1993; Shimek 1996). In northern Patagonia, extensive burning of mesic forests by Euro-American settlers in the 1890s to 1920s resulted in vast areas of even-aged, postfire stands of Austrocedrus chilensis often mixed with Nothofagus dombeyi (Veblen & Lorenz 1987; Kitzberger & Veblen 1999). This regional-scale synchronization of stand age over vast areas has clearly reduced the structural diversity of the landscape. However, the ecological consequences of this landscape homogenization are largely unstudied. By creating more uniform stand ages and structures, the potential for widespread tree mortality from drought or pathogens may have increased. Similarly, these postfire stands present uniform fuel conditions over extensive areas which potentially may increase the hazard of more widespread fire.

Conclusions Changes in fire frequencies at multi-decadal time scales in the Colorado Front Range and in northern Patagonia coincide with changes in human activities as well as climatic variation. Interannual variations in fire regimes closely track regional climatic variability. Although climatic variability overrides human influences on fire regimes at an interannual scale, human activity can be of equal or greater importance in determining fire frequency at multi-decadal scales. This is most evident in the case of 20th century fire exclusion in these conifer woodlands, but humans may also have increased fire occurrence through both intentional and accidental ignitions during the late 19th century. Climatic conditions conducive to widespread fire in both regions result from fuel-enhancing periods of above-average moisture availability followed by one to several years of fuel desiccating drought. In both Colorado and Patagonia, these alternating conditions of above and below average moisture availability are often associated with ENSO events. Long-term trends

205 in fire occurrence parallel long-term trends in interannual variability of tree growth and in ENSO activity. An increase in the frequency of ENSO events during the 19th century probably increased the frequency of alternations between wet and dry periods that are favorable to fire spread. Similar responses of the regional climates and fire regimes to centennial-scale variations in ENSO activity, in combination with similar timing of land use changes, appear to have resulted in an inter-hemispheric synchronization of trends in fire occurrence. These similar trends in fire occurrence have contributed to similar patterns of forest structures, forest health, and hazard of catastrophic wildfire in the Colorado Front Range and northern Patagonia.

Acknowledgements Research was funded by the National Science Foundation of the U.S.A., the National Geographic Society, the Biological Resources Division of the U.S. Geological Survey, the Department of Open Space of the City of Boulder, and the Council for Research and Creative Work of the Graduate School of the University of Colorado. For research assistance and useful discussions of parts of this research we thank L. Daniels, J. Donnegan, D. Lorenz and R. Villalba.

References Arroyo, M. T. K., Cavieres, L., Marticorena, C. & Muñoz, M. 1995. Convergence in mediterranean floras of central Chile and California: insights from comparative biogeography. Pp. 43–88. In: Arroyo, M. T. K, Fox, M. & Zedler, P. (eds), Ecology and biogeography of mediterranean ecoystems in Chile, California, and Australia. Springer-Verlag, New York. Baisan, C. H. & Swetnam, T. W. 1990. Fire history on a desert mountain range: Rincon Mountain Wilderness, Arizona, U.S.A. Can. J. For. Res. 20: 1559–1569. Barden, L. S. 1989. Repeatability in forest gap research: Studies in the Great Smoky Mountains. Ecology 70: 558–559. Barros, V., Cordón, V., Moyano, C, Méndez, R., Forquera, J. & Pizzio, O. 1983. Cartas de precipitación de la zona oeste de las provincias de Rio Negro y Neuquén. Report to the Facultad de Ciencias Agrarias, Universidad Nacional del Comahue, Cinco Saltos, Neuquén, Argentina. Box, E. O. 1981. Macroclimate and plant form. Junk Publishers, The Hague. Brown, P. M., Kaufmann, M. R. & Shepperd, W. D. 1999. Longterm, landscape patterns of past fire events in a montane ponderosa pine forest of central Colorado. Landscape Ecol. 14: 513–532. Buchholtz, C. W. 1983. Rocky Mountain National Park: a history. Colorado Associated University Press, Boulder, Colorado. Clements, G. E. 1910. The life history of lodgepole burn forests. U.S. Dep. Agricult. For. Serv. Bull. 79: 7–56.

Cox, G. 1963. Viajes a las regiones septentrionales de Patagonia 1862–1863. An. Univ. Chile 23: 3–239; 437–509. Daniels. L. D. & Veblen, T. T. 2000. ENSO effects on temperature and precipitation of the Patagonian-Andean region: Implications for biogeography. Phys. Geogr. 21: 223–243. Diaz, H. F. & Kiladis, G. N. 1992. Atmospheric teleconnections associated with the extreme phases of the Southern Oscillation. Pp. 7–28. In: Diaz, H.F. & Markgraf, V. (eds), El Niño: Historical and paleoclimatic aspects of the southern oscillation. Cambridge University Press, Cambridge. Donnegan, J. A. 2000. Climatic and human influences on fire regimes in Pike National Forest. Ph.D. thesis, University of Colorado, Boulder, Colorado, U.S.A. Dunbar, R., Wellington, G. M., Colgan, M. W. & Glynn, P. W. 1994. Eastern Pacific sea surface temperature since 1600 A.D.: The d18 O record of climate variability in Galapagos corals. Paleoceanography 9: 291–316. Fonck, F. 1896. Libro de los diarios de Fray Francisco Menéndez. C.F. Niemeyer, Valparaiso, Chile. Goldblum, D. & Veblen, T. T. 1992. Fire history of a ponderosa pine/Douglas-fir forest in the Colorado Front Range. Phys. Geogr. 13: 133–148. Grau, H. R. & Veblen, T. T. 2000. Rainfall variability, fire, and vegetation dynamics in subtropical montane ecosystems in northwestern Argentina. J. Biogeogr. 27: 1107–1121. Graybill, D. A. 1992. Condition of coniferous forests of the Colorado Front Range. Part A: Mixed species in unmanaged oldgrowth stands. Pp. 370–384. In: Olson, R. K. & Binkley, D. & Böhm, M. (eds), The response of western forests to air pollution. Springer-Verlag, New York. Greenland, D., Burbank, J. Key, J., Klinger, L., Moorhouse, J., Oaks, S. & Shankman, D. 1985. The bioclimates of the Colorado Front Range. Mount. Res. Develop. 5: 251–262. Grissino-Mayer, H. D. 1995. Tree-ring reconstructions of climate and fire history at El Malpais National Monument, New Mexico. Thesis. University of Arizona, Tucson, Arizona. Grissino-Mayer, H. D. & Swetnam, T. W. 2000. Century-scale climate forcing of fire regimes in the American Southwest. Holocene 10: 213–220. Hadley, K. S. & Veblen, T. T. 1993. Stand response to western spruce budworm and Douglas-fir bark beetle outbreaks, Colorado Front Range. Can. J. For. Res. 23: 479–491. Hafen, L. R. & Hafen, A. W. 1956. Rufus B. Sage, his letters and papers 1836–1847. Volume 2. The Arthur H. Clark Company, Glendale, California. Huggard, D. J. & Arsenault, A. 1999. Comment– Reverse cumulative standing age distributions in fire-frequency analysis. Can. J. For. Res. 29: 1449–1456. Jack, J. G. 1899. Pikes Peak, Plum Creek, and South Platte Reserves. Twentieth Annual Report of the United States Geological Survey to the Secretary of the Interior, 1898–1899. United States Government Printing Office, Washington, D.C. Kiladis, G. N. & Diaz, H. F. 1989. Global climatic anomalies associated with extremes in the Southern Oscillation. J. Clim. 2: 1069–90. Kipfmueller, K. F. & Baker, W. L. 1998. Fires and dwarf mistletoe in a Rocky Mountain lodgepole pine ecosystem. For. Ecol. Manag. 108: 77–84. Kitzberger, T., Steinaker, D. F. & Veblen, T. T. 2000. Establishment of Austrocedrus chilensis in patagonian forest-steppe ecotones: facilitation and climatic variability. Ecology 81: 1914–1924. Kitzberger, T., Swetnam, T. W., & Veblen, T. T. 2001. Interhemispheric synchrony of forest fires and the El Niño-Southern Oscillation. Glob. Ecol. Biogeogr. 10: 315–326.

206 Kitzberger, T. & Veblen, T. T. 1997. Influences of humans and ENSO on fire history of Austrocedrus chilensis woodlands in northern Patagonia, Argentina. Ecosc. 4: 508–520. Kitzberger, T. & Veblen, T. T. 1999. Fire-induced changes in northern Patagonian landscapes. Landscape Ecol. 14: 1–15. Kitzberger, T., Veblen, T. T. & Villalba, R. 1997. Climatic influences on fire regimes along a rain forest-to-xeric woodland gradient in northern Patagonia, Argentina. J. Biogeogr. 24: 35–47. Laven, R. D., Omi, P. N., Wyant, J. G. & Pinkerton, A. S. 1980. Interpretation of the fire scar data from a ponderosa pine ecosystem in the central Rocky Mountains, Colorado. United States Department of Agriculture Forest Service, General Technical Report RM-81. Markgraf, V. & Anderson, L. 1994. Fire history of Patagonia: Climate versus Human Cause. Revista do Instituto Geográfico do Sao Paulo 15: 33–47. Marr, J. W. 1961. Ecosystems of the east slope of the Front Range in Colorado. University of Colorado Studies Series in Biology 8. Martonne, E. de. 1926. Une nouvelle fonction climatologique: L’indice d’aridité. Météorologie 2: 449–458. Mast, J. N., Veblen, T. T., & Linhart, Y. B. 1998. Disturbance and climatic influences on age structure of ponderosa pine at the pine/grassland ecotone, Colorado Front Range. J. Biogeogr. 25: 743–755. Mast, J. N., Veblen, T. T. & Hodgson, M. E. 1997. Tree invasion within a pine/grassland ecotone: An approach with historic aerial photography and GIS modeling. For. Ecol. Manag. 93: 187–194. Michaelsen, J. & Thompson, L. G. 1992. A comparison of proxy records of El Niño/ Southern Oscillation. Pp. 323–348. In: Diaz, H. F. & Markgraf, V. (eds), El Niño: Historical and paleoclimatic aspects of the Southern Oscillation. Cambridge University Press, Cambridge. Mooney, H. A. (ed.) 1977. Convergent evolution in Chile and California: Mediterranean climate ecosystems. Dowden, Hutchinson & Ross, Stroudsburg, Pennsylvania. Mooney, C. Z. & Duval, R. D. 1993. Bootstrapping: a nonparametric approach to statistical inference. Sage University Paper Series on Quantitative Applications to Social Sciences 95. Newbury Park, California. Quinn, W. H. 1992. A study of Southern Oscillation-related climatic activity for A.D. 622–1990 incorporating Nile River flow data. Pp. 119–149. In: Diaz, H. F. & Markgraf, V. (eds), El Niño: Historical and paleoclimatic aspects of the southern oscillation. Cambridge University Press, Cambridge. Read, J. & Brown, M. J. 1996. Ecology of Australian Nothofagus forests. Pp. 131–181. In: T. T. Veblen, Hill R. S. & Read, J. (eds), The ecology and biogeography of Nothofagus Forests. Yale University Press, New Haven, Connecticut. Read, J. & Hill, R. S. 1985. Dynamics of Nothofagus-dominated rainforest on mainland Australia and lowland Tasmania. Vegetatio 63: 67–78. Relva, M. A. & Veblen, T. T. 1998. Impacts of introduced large herbivores on Austrocedrus chilensis forests in northern Patagonia, Argentina. For. Ecol. Manag. 108: 27–40. Ropelewski, C. F. & Jones, P. D. 1987. An extension of the TahitiDarwin Southern Oscillation index. Monthly Weather Rev. 115: 2161–2165. Rothkugel, M. 1916. Los bosques patagónicos. Ministerio de Agricultura, Buenos Aires, Argentina. Rowdabaugh, K. M. 1978. The role of fire in the ponderosa pine-mixed conifer ecosystem. Masters thesis, Colorado State University, Fort Collins, Colorado, USA. Shimek, S. W. 1996. A dendrochronological history of western spruce budworm (Choristoneura occidentalis Freeman) infes-

tations and outbreaks in the northern Colorado Front Range. Masters thesis. University of Colorado, Boulder, Colorado, USA. Shinneman, D. J. & Baker, W. L. 1997. Nonequilibrium dynamics between catastrophic disturbances and old-growth forests in ponderosa pine landscapes of the Black Hills. Conserv. Biol. 11: 1276–1288. Skinner, T. V. & Laven, R. D. 1983. A fire history of the Longs Peak region of Rocky Mountain National Park. Pp. 71–74. In: Proceedings of the seventh conference on fire and forest meteorology, April 25–28, Fort Collins, CO. American Meteorological Society, Boston, MA, U.S.A. Smith, P. 1981. A look at Boulder from settlement to city. Pruett, Boulder, Colorado, U.S.A. Sudworth, G. B. 1899. White River Plateau Timber Land Reserve. Pp. 117–179. In: Twentieth Annual Report of the United States Geological Survey, Part V. Forest Reserves. United States Government Printing Office, Washington, D.C., U.S.A. Stahle, D. W., D ’Arrigo, R. D., Krusic, P. J., Cleaveland, M. K., Cook, E. R., Allan, R. J., Cole, J. E., Dunbar, R. B., Therrell, M. D., Gay, D. A., Moore, M. D., Stokes, M. A., Burns, B. T., Villanueva-Diaz, J. & Thompson, L. G. 1998. Experimental dendroclimatic reconstruction of the Southern Oscillation. Bull. Am. Meteorol. Soc. 79: 2137–2152. Swetnam, T. W. & Betancourt, J. L. 1998. Mesoscale disturbance and ecological response to decadal climatic variability in the American Southwest. J. Clim. 11: 3128–3147. Swetnam, T. W. & Lynch, A. M. 1989. A tree-ring reconstruction of western spruce budworm history in the southern Rocky Mountains. For. Sci. 35: 962–986. Tice, J. H. 1872. Over the plains and on the mountains. Industrial Age Printing, St. Louis, Missouri, U.S.A. Tsoni, A. A., Roebber, P. J. & Elsner, J. 1998. A characteristic time scale in the global temperature record. Geophys. Res. Lett. 25: 2821–2823. Veblen, T. T. & Alaback, P. B. 1996. A comparative review of forest dynamics and disturbance in the temperate rain forests of North and South America. Pp. 173–213. In: Lawford, R. G., Alaback, P. B. & Fuentes, E. R. (eds), High latitude rain forests and associated ecosystems of the west coast of the Americas: Climate, hydrology, ecology and conservation. Springer-Verlag, New York. Veblen, T. T., Donoso, C., Kitzberger, T. & Rebertus, A. J. 1996. Ecology of southern Chilean and Argentinean Nothofagus forests. Pp. 293–353. In: Veblen, T. T., Hill R. S. & Read, J. (eds), The ecology and biogeography of Nothofagus Forests. Yale University Press, New Haven, Connecticut. Veblen, T. T., Donoso, C., Schlegel, F. M. & Escobar, B. 1981. Forest dynamics in south-central Chile. J. Biogeogr. 8: 211–247. Veblen, T. T., Kitzberger, T. & Donnegan, J. 2000. Climatic and human influences on fire regimes in ponderosa pine forests in the Colorado Front Range. Ecol. Applic. 10: 1178–1195. Veblen, T. T., Kitzberger, T., Villalba, R. & Donnegan, J. 1999. Fire history in northern Patagonia: the roles of humans and climatic variation. Ecol. Monogr. 69: 47–67. Veblen, T. T. & Lorenz, D. C. 1986. Anthropogenic disturbance and recovery patterns in montane forests, Colorado Front Range. Phys. Geogr. 7: 1–24. Veblen, T. T. & Lorenz, D. C. 1987. Post-fire stand development of Austrocedrus-Nothofagus forests in Patagonia. Vegetatio 73: 113–126. Veblen, T. T. & Lorenz, D. C. 1988. Recent vegetation changes along the forest/steppe ecotone in northern Patagonia. Ann. Asso. Am. Geogr. 78: 93–111.

207 Veblen, T. T. & Lorenz, D. C. 1991. The Colorado Front Range: A century of ecological change. University of Utah Press, Salt Lake City, Utah, U.S.A. Veblen, T. T., Mermoz, M., Martín, C., & Kitzberger, T. 1992. Ecological impacts of introduced animals in Nahuel Huapi National Park, Argentina. Conserv. Biol. 6: 71–83. Villalba, R. 1994. Tree-ring and glacial evidence for the Medieval Warm Epoch and the Little Ice Age in southern South America. Clim. Change 26: 183–197. Villalba, R., Cook, E. R., D’Arrigo, R. D., Jacoby, G. C., Jones, P. D., Salinger, M. J. & Palmer, J. 1997. Sea-level pressure variability around Antarctica since A.D. 1750 inferred from subantarctic tree-ring records. Clim. Dynamics 13: 375–390. Villalba, R. & Veblen, T. T. 1997a. Regional patterns of tree population age structures in northern Patagonia: climatic and disturbance influences. J. Ecol. 85: 113–124.

Villalba, R. & Veblen, T. T. 1997b. Spatial and temporal variation in tree growth along the forest-steppe ecotone in northern Patagonia. Can. J. For. Res. 27: 580–597. Villalba, R. & Veblen, T. T. 1998. Influences of large-scale climatic variability on episodic mortality at the forest-steppe ecotone in northern Patagonia. Ecology 79: 2624–2640. Willis, B. 1914. El norte de la Patagonia. Dirección de Parques Nacionales, Buenos Aires, Argentina. Zimmerman, G. T. & Laven, R. D. 1984. Ecological implications of dwarf mistletoe and fire in lodgepole pine forests. Pp. 123– 131. In: Hawksworth, G. G. & Scharpf, R. F. (eds), Proceedings of the symposium on the biology of dwarf mistletoes. U.S. Dep. Agricul. For. Serv. Gen. Techn. Rep. RM-111.