This file was created by scanning the printed publication. Errors identified by the software have been corrected; however, some errors may remain.

Fire History and in the Southwestern United States1

Thomas W. Swetnam2

In the study of centuries-long pat­ ment lexicon, it was clear that the challenges. The focus is on historical terns of change in forest pendulum had moved back toward records of fire and climate in Arizona one can observe recurrent themes, center. and New during the last 300 even "cycles," if we define this term Part of the cycling of fire policy years. Analyses of weather records in a nonclassical sense. For example, may be linked to the cycling of fire also reveal that one particular cli­ consider the well known successional regimes. Public (i.e., political) sup­ matic phenomena, the El Nino­ pathway of aspen stands establishing port of the early total suppression Southern Oscillation (ENSO), seems after stand-replacing fires in mixed­ policy grew out of the shock of dev­ to explain a substantial amount of conifer or spruce-fir forests. The as­ astating fires in the late 1800s, such the year-to-year fire load in the pen are, in tum, replaced by regener­ as_the Hinckley and Peshtigo fires in southwest. The ENSO-Southwestem ated shade-tolerant conifers and, fi­ the Lake States, and the 1910 fires of fire teleconnection (Swetnam and nally, recurrence of another fire the Northern Rockies. Now, follow­ Betancourt, in preparation) suggests storm may repeat the process. Man­ ing the catastrophic fires in Califor­ that during some years it may be agement practices on public nia and the last possible to forecast fire season sever­ also seem to run in cycles. Consider year, and the Yellowstone and other ity several months in advance. the historical fact that fire suppres­ western fires this year, we are seeing sion was essentially nonexistent in a similar "fire shock" both in the western forests before the first dec­ public and management agen­ Methods ade of this century. "Let burn" prac­ cies. Are we now poised for another tices held sway in a number of areas shift in fire policy, presumably back Historical fire records were com­ in the 1910s and 20s, particularly toward the total suppression side? I piled from two general sources: (1) California. The "hit-em hard and hit­ think not, but any reassessment of Forest Service documents covering em fast" strategy of the 10 AM policy policy has much to gain from a his­ the post-1900 period, and (2) fire-scar finally won out completely in a torical perspective. If we are to avoid records on trees covering the pre- heated controversy in the early further "fire shocks" we must recog­ 1900 period. The Forest Service docu­ 1930s, and the pendulum was then nize that regional-scale fire years ments included data on number of solidly on the total suppression side. have occurred in the past, sometimes fires and acres burned per year in the In the 1970s and 80s land manage­ despite aggressive total suppression Southwestern (Arizona and ment agencies in the western U.S. policies, and these events will very New Mexico), USDA Forest Service. shifted back toward the approach of likely recur in the future. Manage­ These data were analyzed by size using fire for management purposes. ment strategies must be geared to class and totals. Sources were annual AI though the term "let-burn" was minimizing these resource impacts National Forest Fire Reports (USDA, still banned from the fire manage- through forecasting of hazardous Forest Service) and a detailed study conditions, fire-fighting readiness, of southwestern lightning fire data 1 Panel paper presented at the confer­ and prudence in the use of pre­ conducted by Barrows (1978). A ence, Effects of Fire in Managementof scribed fire. - Southwestern Natural Resources (Tucson, longer record was compiled for the Al, November 14-17, 1988). The main theme of this paper is Gila National Forest in southwestern 2 Assistant Professor of Dendrochronol­ that historical and ecological knowl­ New Mexico from fire atlases and ogy, Laboratory of Tree-Ring Research, Uni­ edge can provide managers with·per­ individual fire reports (Swetnam versity of Arizona, Tucson, AZ 8572 7. spectives necessary to meet these 1983a), and from an early fire statis-

6 tics study by E. W. Loveridge in 1926 and Swetnam 1984). Table 1 lists guards on horseback available for (files at Gila National Forest Supervi­ published and unpublished sources fire fighting from about 1909 to 1933, sor's Office, Silver , NM). of the data sets and various other and the number of fires reported was Pre-1900 fire occurrence data were characteristics of the fire regimes. relatively low during this period compiled from fifteen fire-scar chro­ with no noticeable upward trend nologies. These include 13 chronolo­ (fig. 2). Then in 1934 over 100 men of gies from Arizona and New Mexico, Results the Civilian Conservation one from Mexico (Sierra Los Ajos, (CCC) were stationed within the Na­ Sonora) and another from Post-1900 Fire History tional Forest and the fire record (Guadalupe Mountains National shows a obvious upward trend. Park) (fig. 1.). The fire-scar chro­ A positive trend in number of fires Other detection and fire suppression nologies were developed by collect­ reported each year was evident in improvements also correspond to ing full or partial cross the records of fire occurrence. Most increases in total numbers of re­ samples from living and dead fire­ of this effect was due to improve­ ported fires (fig. 2). Most of this in­ scarred trees, crossdating the annual ments in detection and fire-fighting crease, however, was in the number rings (Stokes and Smiley 1968) and resources available to send to fires. of small fires (class A, 0.25 acres or observing the position of fire scars For example, in the Gila National less). It seems likely that in the first within the annual rings (Dieterich Forest there were only 4 to 6 fire 20-year period the Gila fire guards


_____ ... ----·-·; {-\ . \ I 'I ' '\ I ~ I

\ 'I . 't


Figure 1.-National Forests In the Southwestern Region, U.S. Forest Service, Arizona and New Mexico. Fire scar collections are numbered and are described in table 1.

7 were probably unable to detect all 1971 and 1974; 1979 was also an un­ rise in area burned per lightning­ fires that started, and even if they usually heavy fire-load year. caused fire from 1940 to at least 1979 had they would not have been able to One possible bias in the record af­ seems to be a genuine phenomenon go to all of them. ter about 1974 was the increased area that suggests a worsening fire situ­ Although improvements in detec­ burned under prescribed natural fire ation in the Southwestern Region. tion and increase in manpower ex­ programs, where lightning ignited Obvious trends in total area most of the increase in fire fires were allowed to burn during burned per year were generally not numbers in the Gila, there also ap­ special conditions. However, when observed (fig. 4). However, large pears to be more larger fires. The the area burned under this program peaks during certain years of this number of large fires reported in the in the Gila Wilderness (which had record are notable, reflecting espe­ Southwestern Region may also have the most active program well into the cially severe fire years (e.g., 1925, a slightly upward trend (fig. 3). Table 1980s) was discounted from the 1970- 1946, 1951, 1956, 1959, 1960, 1971, 2 compares the number of fires by 1979 period, the average area burned 1974, 1979, and 1985). In some years, size class for selected early and late per lightning-caused fire was still large burned area values can be periods in the Southwestern Region 6.66 ha, which was more than 40% traced to one or a few very large fires and the Gila. For the Southwestern greater than the previous 10-year pe­ on a single National Forest, such as Region, only the number of class A riod. The magnitude of the pre­ 1951 when two fires burned over fires has significantly increased in the scribed natural fire effect on area 20,000 ha in the Gila. However, the modern period; however, the num­ burned per fire for the 1980-1987 pe­ majority of the peaks in fire activity ber of fires in all size classes has in­ riod has not yet been assessed be­ in the Southwestern Region record creased in the Gila. cause of lack of data. In any case, the represent regional-scale fire years While improved detection and fire-fighting resources may account for the increase in number of small fires the increase in number of larger fires, may have a different explana­ tion. Possibilities include change in climate and/ or change in fuel load­ ings toward conditions favoring lar­ ger fires. Another possible contribut­ ing cause may be increased commer­ cial and public use and access to Na­ tional Forest lands, because there was also a statistically significant in­ crease in average annual number of person-caused fires in the modem period (Gila record: 1909-1939, X= 12; 1960-1986, X= 18, t = 2.87, p = 0.006) (also see fig. 2). It is not known how much of the increase in large fires in the Gila was due to ignitions because data on per­ son-caused fires classified by size were not obtained. Other evidence suggests that an increasing trend in fire load was not related to person-caused fires. Con­ sider the lightning fire data for the Southwestern Region (table 3). Barrows (1978) compiled the first 36 years of this record, and he also noted the steady rise in area burned per fire. He pointed out that the high area burned figure for the 1970s was strongly influenced by large fires in

8 or very soon after the turn of the cen­ 700 Helicopters and ~omber1 tury. Very few or no fire scars were recorded on any of the trees repre­ 600 sented in figure 5 after 1900, with the exception of the chronology from Si­ rll 500 erra Los Ajos. This chronology re­ QJ corded a relatively uninterrupted re­ .~ gime of frequent surface fires extend­ ~ oiOO ing into the 1980s when the trees Smokejumpers ~ were sampled. Recurring fires in the 0 . :'j00 twentieth century at this site may re­ 0 flect few or no attempts to suppress z 200 t fires on this relatively remote moun­ Lightning-caused taintop of. . 100 Many fires during the presuppres­ sion era burned for months at a time, and they covered thousands of hec­ 1919 1929 lQ39 1949 1959 1969 1979 tares. We know this must be true be­ Year cause lightning ignites fires in some ponderosa pine forests as early as Figure 2.-Number of fires reported per year, 1909-1986, In the Gila National Forest, New April, with at least two months of Mexico. Approximate timing of Improvements In fire-fighting technology correspond with Increases In numbers of reported fires. dry conditions remaining. Some fires might have persisted through the when large fires were reported in OJl$trates the frequent and aperiodic rainy season of July and August and many of the National Forests. nature of forest fires in ponderosa flared up again during drier periods pine and mixed-conifer forests of the of late summer and early fall. Also, Southwestern Region during the two fire scar records collected from nu­ Pre-1900 Fire History centuries before 1900. One of the merous trees scattered over ponder­ most striking and consistent observa­ osa pine forests as large as 2,000 ha Time series of the fifteen fire-scar tions derived from nearly all of these show matching fire dates among chronologies listed in table 1 are il­ histories was the sudden end of a most trees for most fires, indicating lustrated in figure 5. This chart dem- surface fire regime in the late 1800s, very large burns were typical in this type. The mean fire intervals for all 18 chronologies range between approxi­ j mately 2 and 10 years (table 1). Mean 16 -·Class D ~ fire intervals in ponderosa pine sites rn QJ 14 on the Mogollon Rim (central Ari­ J-4 zona and west-central New Mexico) ·~ -Class E+ I ~ 12 1\ and southward range from 4 to 5 II years, while mean fire intervals at 'Cl 10 \ higher elevations in mixed-conifer I types range from 6 to 10 years. The ~ 8 \ (ll mixed-conifer chronologies often re­ ,.c I 8 corded smaller fires at short inter­ E vals. When only larger fires in this :=;j 4 type were considered, the mean fire z intervals were on the order of 15 to 2 25 years (e.g., Dieterich 1983). Three 0 of the northern Arizona chronologies 1940 1945 1950 1955 1980 1965 1970 1975 (Limestone Flats, Chimney Spring Year and Battle Flat) recorded the highest fire frequencies of any forest type, with mean fire intervals ranging be­ Figure 3.-Number of class D (100 to 299 acres) and class E and larger(> 300 acres) light­ tween 1.9 and 2.9 years. Consecutive- ning fires In the Southwestern Region (Barrows 1978). 9 year fires (1-year intervals) have also more than 80% of sampled trees re­ western coast of the . The been noted on rare occasions in some corded fires during the regional-fire Southern Oscillation is measured as of these sites. The higher fire fre­ years. the normalized differences in quencies in northern Arizona chro­ monthly mean pressure anomalies nologies, with exception of the between Tahiti, French and Chuska Mountains, is clearly visible Southwestern Fire and ENSO Darwin, . El Nino and the in figure 5. Events Southern Oscillation are linked in a The mean recurrence intervals global climate complex of changing shown in table 1 are primarily useful El Nino-Southern Oscillation currents, ocean temperatures, for comparative purposes. It is most (ENSO) events are global-scale cli­ atmospheric pressure and tempera­ important to consider the historical matic anomalies that recur at inter­ ture gradients. Subcontinetal or re­ range and variability of fire occur­ vals of 2 to 10 years and at varying gional-scale climatic effects of ENSO rence. One example of temporal vari­ intensities (Philander 1983). TheEl events are highly variable. In some ability can be seen as an apparent Nino pattern is characterized by cases the effects are consistent and shift in fire regimes to longer fire­ weak tradewinds and appearance of extreme, leading to droughts in some free periods beginning in the early high -surface temperatures off the and flooding in others. and middle 1800s in many of the chronologies (fig. 5). A puzzling gap with few or no fires during a period from about 1820 to 1840 is also no­ ticeable in some of the chronologies. This hiatus may be due to wetter conditions as evidenced by climate reconstructions from tree rings; the 1830s-40s period was apparently one of the wettest in the last two hundred years (Schulman 1956, Fritts 1986). Also, reduction in fire frequency dur­ ing this period may have been associ­ ated with intensified sheep grazing in New Mexico during the 1820s (Denevan 1967). Intensive grazing removes fine grassy fuels important in spreading fires. However, infor­ mation about mid-19th century stock numbers in southern New Mexico and Arizona is sketchy. The grazing hypothesis may be supported by the fact that many of the fire chronolo­ gies recorded a final fire in the late 1800s around the beginning of an in­ tensive grazing era, and 20 years or more before the advent of organized fire suppression efforts by the USDA Forest Service. The fire scar chronologies also show a remarkable correspondence of fire years across the Southwest (fig. 5). These years are also evident in a fire area index (fig. 6). The fire area index reveals that regional-fire years (labeled years) were often large fires within the sites where they were recorded. The individual chronolo­ gies (fig. 5) show that in some cases

10 A possible link between ENSO re­ 1988, Betancourt 1988). This condi­ the Southwest until the fall and win­ lated droughts in the western Pacific tion is due to southward displace­ ter of 1925-26. Note that one of the and fire is evidenced by massive ment of the jet stream and westerlies most severe ENSO events of the past bush fires in Australia during the in the eastern Pacific and devel­ two centuries occurred in 1982 and 1939 and 1982-83 El Nino events. opment of tropical storms over the 1983, when the area burned in the Also, one of the largest forest fires in warming off the coast of Baja, Southwest was the second and third history occurred in the tropical for­ California. Subsequent anti-cyclonic lowest over the past 50 years. The ests of the Kalimantan of movement of these storms into lowest area burned was in the year Indonesia in 1982. This gigantic fire is northern Mexico, southern Arizona 1941, another very severe ENSO estimated to have burned over 3.1 and New Mexico brings an unsea­ event.. million hectares! sonable influx of moisture to the re­ Initial observations of number of Simard et al. (1985) compared gion during the normally arid spring fires recorded per year in relation to ENSO events to 53-year fire records and fore-summer. This is also the ENSO events has yielded inconsis­ from large regional areas of the critical burning period in the South­ tent results. Simard et al. (1985) also and found that only the western Region (fig. 7). noted that these types of data were southeastern region had a statisti­ Figure 8 shows the timing of less clearly related to ENSO than cally significant relationship (inverse) ENSO events as classified by Quinn area burned. This may be due to the with this climate phenomena. In their et al. (1987) and area burned per year confounding effect of changing fire­ study, Arizona and New Mexico in the Gila National Forest and the detection and fire-fighting resources. were grouped with all the Rocky Southwestern Region. Differences in A more complex relationship may Mountain and states mean hectares burned per year be­ also exist with thunderstorm activity northward to the Canadian border. tween ENSO and non-ENSO years and ENSO. For example, increased They noted, however, that this com­ were evaluated using the Mann­ moisture during some ENSO years parison was probably at too coarse a Whitney test (table 4). The ENSO may actually result in increased scale, and that smaller and more de­ event in 1925 appears to be the only thunderstorm activity which triggers tailed regional studies may identify outstanding exception to a consistent more lightning caused fires, but stronger relationships. correspondence of ENSO events and lower overall area is burned because In the southwestern United States, the lowest area burned years (fig. 8). of higher fuel moisture. Weak or ENSO events are most consistently Unlike most ENSO events, the spring moderate El Nino events also seem related to wetter than average spring of 1925 was actually quite dry, and to be less clearly associated with re­ and fall seasons (Andrade and Sellers wetter conditions did not develop in duced southwestern fire activity than severe or very severe events. Again, it seems possible that ignition rates and fuel moisture may be involved. Gila National Forest. rn 20000 Weak or moderate El Nino events Cl) 1909-1966 ~ may deliver enough moisture to the ~ m Southwest to increase the level of ~ 0 CJ 10000 ~ thunderstorm activity but not suffi­ QJ cient moisture to reduce fire spread. ::r:= The next ENSO fire comparison utilized the much longer-term per­ spective of fire-scar chronologies. 40000 The two--century record of fire occur­ Southwestern Region. rence derived from the five Gila fire­ ~ 30000 1940-1986 J...t scar chronologies was compared to !}20000 severe and very severe ENSO events 0 in figure 9. El Nino events for the ~ 10000 period prior to the mid-nineteenth century were reconstructed from his­ o~~~~~~~~~~~~ 1909 1919 1929 1939 1949 1959 1969 1979 toric records of droughts, floods, fluctuations in fisheries off the South -Years coast, and other variables that are known to be closely linked to Figure 4.-Area burned per year in Gila National Forest and Southwestern Region. The Gila data include person and lightning-caused fires, while Southwestern Region data Include the El Nino pattern (Quinn et al. only lightning fires. fire years are labeled. 1987).

11 With only a few exceptions the se­ vere ENSO events corresponded to f Chuska Mountain• years of reduced fire activity as I h. .J I ,J&.J.J.. I I I h I ,, "' I • L I. measured by fire-scarred trees (fig. IJ... J l. 9). Another fairly consistent pattern was occurrence of large fire years one or two years following severe ENSO events (fig. 9). ~ BalU. Fl&l ·JJJ1 LJ!L.LJ ••• 1, h,.lluJ., ll, ,I Discussion: Implications for Fire Management J IJ.me•ton• Flat. I I t IIJ ~.Lil I 1.1114 I. "IL.t.J.J.hiJLJJI,,I I u I l Trends and Extreme Events .P'rtjoles Yesa j Two particular warnings to man­ I I_ 1 J _ I I j II ,I .... I &l I agers emerge from the study of fire records. The first is a trend of in­ creasing fire load. Evidence of this I trend includes higher numbers of I large fires in the Gila National Forest Cllit.a RJ.dae in the modern period and larger area I I J II Ill 1... I I, II I t I ,II II I burned per lightning-caused fire in I-II I. d II the shorter record from the entire !Boorwallow Southwestern Region. This trend !U I I. 1 I I I II II I may indicate that fire control will be I I I I I an increasingly difficult task in com­ ing years. Possible causes of this Langstralh w... I I I I trend may be climate change, man­ III II, I II " II ,, I ~ "' I I. I II I~ I I agement practices, or both. Management practices might pro­ f llcl

12 regime of southwestern forests. This tury and the fire-scar record extend­ Climatic warming is another con­ is not particularly new information ing back to 1700, it appears that at cern. Regardless of whether or not a to southwestern fire-fighting veter­ least 3 or 4 regional-fire years oc­ warming effect due to increased C02 ans who remember, for example, the curred per century. The occurrence in the atmosphere is already reflected 1956, 1971 and 1974 fire seasons. of extreme regional-fire years sug­ in global weather patterns, there is a Nevertheless, consideration of his­ gests that atmospheric conditions building consensus among atmos­ tory extending beyond the reach of were responsible, because these are pheric scientists that such an effect, human memories can serve to verify the only environmental variables that at some unknown intensity, is virtu­ the existence of this phenomenon, can account for such spatially large­ ally imminent. Since some of the pro­ and possibly help identify the proba­ scale phenomena. It is also likely that jected warming scenarios show in­ bility of its recurrence. these conditions were at least partly creasing droughtiness in the south­ One of the remarkable features of cumulative through some period of west (Schlesinger and Mitchell 1985) the two-century southwestern fire time preceding extreme fire events, this effect may also contribute to a history represented in the fire-scar and therefore some ad vance warning worsening fire situation. chronologies was the synchrony of may be present. There is clearly a From an ecological point of view, large fire years among the widely need for further research of historical the variability in fire regimes is more scattered sites. Based on both the climate and fire records to pinpoint likely to be important to plant com­ Forest Service data base for this cen- these conditions. munities than mean values computed for some arbitrary period. For ex­ ample, unusual long periods without > 50% fire may lead to increased establish­ ment of certain species that are intol­ erant of fire during the first years of life. The co-occurrence of such fire­ free periods and wetter climatic con­ ditions may also be extremely impor­ tant to species with episodic regen­ eration patterns, such as ponderosa ~ ~ pine. Thus, while statistical summa­ C":) 1.0 CD- l'-- CD -tiC ries of fire chronologies are useful for CD l"- ~ ~ l'- ..-1 .-4 ...-4 t.'J 1Dr:-- m .... -till or- ~ tQ general comparisons of different for­ rt cc ...... rt ests, fire's influence on the ecosys­ tem is strongly a historical process. Southwestern forests may be more a product of relatively short-term and unusual periods of climate and fire > 10% frequency then average or cumula­ tive measures of these long-term his­ tories.

ENSO and Fire

gj 10 All The ENSO-fire teleconnection in ~ the Southwestern Region is clearly tzj B inverse, i.e., significantly wetter ..... 6 springs and summers during ENSO 0 events results in a very reduced fire load. Additional work has also 0 shown that the opposite pattern of El Z r'100 1720 1'740 1'160 1780 1800 1820 1640 lBBO 1BBO 1900 Nino, sometimes referred to as La Nina (Kerr 1988), seems to often cor­ Figure 6.-Fire-area Index computed as number of sites (fire-scar chronologies) recording respond to peak fire. occurrence years fires per year for the period 1700-1900. Fifteen sites are Included. Fires recorded by any tree within the sites are shown in the lowermost plot, while fires recorded by more than 10, 20, (Swetnam and Betancourt, in prep.). and SOOfo of fire-scar susceptible trees are shown above. Regional fire years are labeled. The most promising application of

13 these findings would be a predictive for several centuries prior to the be­ forest ecosystems must also be tool for anticipating fire season se­ ginning of active forest land manage­ weighed against practical considera­ verity. Work by Bradley et al. (1987) ment and fire suppression. Through tions, such as the generation of and Andrade and Sellers (1988) has time the dynamic effects of surface smoke and the hazard of escaped shown that temperature and precipi­ fires were second in importance only fire. tation response in North American to the change of seasons. Trees that The synchronous occurrence of records usually lags 3 to 6 months germinated, established and grew large fires in many of the widely behind the onset of ENSO events as through the majority of their lifespan scattered forests of the Southwest in measured by sea-surface tempera­ in a fire regime of repeated surface particular years is an outstanding tures and the southern oscillation. fires still comprise the vast majority feature of both pre- and post-1900 Thus, the ability of meteorologists to of harvestable timber products in the fire records. Severe and very severe predict ENSO conditions during the region, as well as contributing to ENSO events appear to be consis­ winter months before a fire season is other forest values. From this per­ tently associated with reduced fire likely to improve in coming years spective of the past, managers must occurrence in the region. Regardless (Barnett et al. 1988). assess the ecological consequences of of whether the ENSO-fire teleconnec­ What are some of the management 80 years of fire suppression in forest tion proves useful for predictive pur­ implications of such predictive types that have adapted to fire over poses, this finding and the observa­ power? One implication would be for many thousands of years. What will tion of regional-scale fire years ar­ the planning and implementation of our forests look like 100 or 200 years gues that, to a large degree, regional prescribed burning plans. Prescribed from now if fire does not play the climate patterns control year-to-year burning during the typical arid role it did for centuries before our fire occurrence. A challenge tore­ spring and foresummer in the South­ intervention? The fire histories docu­ searchers and managers is to recog­ west carries an unknown level of risk mented by fire-scar studies remain nize these patterns and to act in time. due to the uncertain timing of arrival one of the strongest scientific argu­ Recommended actions include con­ of the monsoon pattern later in the ments for incorporating fire into the tinued and increased effort to reduce summer, and the possibility of ex­ management of southwestern forests. accumulated fuel loadings through tended dry and windy conditions in In making management decisions, careful use of fire, and consideration the interim. Developing ENSO condi­ ecological knowledge of factors im­ of climatic patterns in fire manage­ tions during winter months, espe­ portant to the healthy functioning of ment planning. cially those suggesting extreme events, may call for stepping up pre­ scribed burning activities in the Tucson Precipitation Fire Occurrence spring and early summer because the likelihood of drying conditions later DEC in the season would be lower. Like­ NOV wise, if further research shows a OCT strong link between La Nina condi­ tions and high fire occurrence, it SEP would be advisable to curtail spring *..--lllllliiiiiiiiiiiiiiiiiiiiiif AUG and summer burning during such ruL years. Even if the ENSO phenomena JUN • El Nino does not prove to be a useful predic­ MAY tive tool for southwestern fire man­ APR • All Fires agers, the high likelihood that region­ llAR al-fire years will recur in the south­ D Lightnin.j west strongly indicates that consid­ FEB eration be given to drying trends JAN during previous and current seasons. 60 50 40 30 20 10 0 0 10 20 30 40 50 mm Percent

Conclusions Figure 7.-Distribution of fires by month (1940-1975) in comparison to a precipitation record from Tucson, Arizona (1868-1986). Peak fire activity Is during the driest period. Especially Fire occurrence records for pre- critical are the two to three weeks just prior to the "Arizona Monsoon" at the end of June and beginning of July. Wetter months during El Nilio years in the winter, spring and fore­ 1900 periods document the ubiquity summer (significant differences Indicated with an asterisk, Mann-Whitney test, p < 0.05) may of fire in the southwestern landscape explain most of the reduction In area burned per year in the Southwestern Region.

14 References Service, Rocky Mountain Forest and western New Mexico. Journal and Range Experiment Station: of Climatology 8:403-410. Ahlstrand Gary M. 1980. Fire history 4-7. Baisan, Christopher H. 1988a. Fire of a mixed-conifer forest in the Allen, Craig D. 1988. Twentieth cen­ history of Rincon Mountain Wil­ Guadalupe Mountains National tury changes in the of the derness, Saguaro National Monu­ Park. In: Stokes, Marvin A.; Di­ Jemez Mountains landscape, New ment. Final Report to U.S. De­ eterich, John H., tech. coords. Fire Mexico. Abstract in: Bulletin of the partment of Interior, National history workshop: Proceedings of Ecological Society of America. Park Service, Saguaro National a technical conference, 1980 Octo­ 69(2):55. Monument, Tucson. Unpublished ber, Tucson, AZ. Gen. Tech. Rep. Andrade, Edward R.; Sellers, Wil­ report on file at Laboratory of RM-81. Fort Collins, CO: U.S. De­ liam D. 1988. El Nino and its ef­ Tree-Ring Research, University of partment of Agriculture, Forest fect on precipitation in Arizona Arizona, Tucson. 82 p. Baisan, Christopher H. 1988b. Fire 100000 history of a mixed conifer forest, - Gila National Forest. Black Mountain, Gila Wilderness. t Final Report to U.S. Department -+ Southwestern Region of Agriculture, Forest Service. 10000 -tW Unpublished report on file at 0 Laboratory of Tree-Ring Research, r--4 192 University of Arizona, Tucson. 00 10 p. (1,) 1000 ~ Barnett, T.; Graham, N.; Cane, M.; cd Zebiak, S.; Dolan, S.; O'Brien, D.; +l c.> Legler, D. 1988. On the prediction Q) of the El Nino of 1986-1987. Sci­ ~ ence 241:192-196. Barrows, JackS. 1978. Lightning fires in southwestern forests. Final Report to U.S. Department of Ag­ 1915 1Q25 1935 1945 1955 1985 19?5 1985 riculture, Forest Service, Inter­ Year mountain Forest and Range Ex­ periment Station, under coopera­ Figure 8.-Area burned per year (logarithmic scale) in Southwestern Region and Gila Na­ tive agreement 16-568-CA with tional Forest in relation to moderate and severe El Nino events. Severe or very severe events according to Quinn et al. (1987) occurred in 1911-1912, 1914, 1917, 1925-1926, 1932, 1940- Rocky Mountain Forest and Range 1941, 1957-1958, 1972-1973, 1982- 1983. All other labeled dates were moderate El Nino Experiment Station, Fort Collins, CO. Department of Forest and Wood Science, Colorado State University, Fort Collins. 154 p. Betancourt, Julio L. 1988. El Nino/ Southern Oscillation (ENSO) and climate of the southwestern U.S. In: Proceedings of Paleoclimate Workshop, National Science Foun­ dation and NOAA, 1988 February, Boston, MA. Bradley, R. S.; Diaz, H. F.; Kiladis, G. N.; Eischeid, J. K. 1987. ENSO sig­ nal in continental temperature and precipitation records. Nature 327: 497-501. Caprio, Anthony C.; Baisan, Christo­ pher H.; Brown, Peter M.; Swet­ nam, Thomas W. 1988. Fire scar dates fron1 Bandelier National Monument, New Mexico. Final

15 report to U.S. Department of Inte­ Kerr, Richard A. 1988. La Nina's big Schlesinger, M. E.; Mitchell, F. F. rior, Bandelier National Monu­ chill replaces El Nino. Science 241: 1985. Model predictions of the ment, under contract PX 7120-8- 1037-1038. equillibrium climatic response to 0072. Unpublished report on file at Philander, S. G. H. 1983. El Nino increased carbon dioxide. In: Mac­ Laboratory of Tree-Ring Research, Southern Oscillation phenomena. Cracken, M. C.; Luther, F. M., eds. Univ. of Arizona, Tucson. 49 p. Nature 302:295-301. Projecting the climatic effects of Denevan, William M. 1967. Livestock Quinn, William H.; Neal, Victor T.; increasing carbon dioxide, DOE/ numbers in nineteenth-century Antunez de Mayolo, Santiago E. ER- 0237, U.S. Department of En­ New Mexico, and the problem of 1987. El Nino occurrences over the ergy, Washington DC. 83-147. gullying in the Southwest. Annals past four and a half centuries. Schulman, Edmund. 1956. Dendrocli­ of Association of American Geog­ Journal of Geophysical Research matic changes in semiarid Amer­ raphy. 57(4):691-703. 92(C13):14,449-14,461. ica. University of Arizona Press, Dieterich, John H. 1980a. Chimney Romme, William. 1980. Fire History Tucson. 142 p. Spring forest fire history. Gen. Terminology: Report of the Ad Simard, Albert J .; Haines, Donald A.; Tech. Rep. RM-220. Fort Collins, Hoc Committee. In: Stokes, Main, William A. 1985. Relations CO: U.S. Department of Agricul­ Marvin A.; Dieterich, John H., between El Nino/Southern Oscil­ ture, Forest Service, Rocky Moun­ tech. coords. Fire history work­ lation anomalies and wildland fire tain Forest and Range Experiment shop: Proceedings of a technical activity in the United States. Agri­ Station. 8 p. conference, 1980 October, Tucson, cultural and Forest Meteorology Dieterich, John H. 1980b. The com­ Az. Gen. Tech. Rep. RM-81. Fort 36:93-104. posite fire interval-a tool for Collins, CO: U.S. Department of Stokes, Marvin A.; Smiley, Terah L. more accurate interpretation of Agriculture, Forest Service, Rocky 1968. An introduction to tree-ring fire history. In: Stokes, Marvin A.; Mountain Forest and Range Ex­ dating. University of Chicago Dieterich, John H., tech. coords. periment Station: 135-137. Press, Chicago. 73 p. Fire history workshop: Proceed­ Savage, Melissa. 1988. Changing Swetnam, Thomas W. 1983a. Fire his­ ings of a technical conference, 1980 landscape patterns in a managed tory of the Gila Wilderness, New October, Tucson, AZ. Gen. Tech. pine forest. In: Roberts, Rose Mexico. Tucson: University of Rep. RM-81. Fort Collins, CO: U.S. Fabia; Rainman, Kermit; compil­ Arizona. 143 p. M.S. thesis. Department of Agriculture, Forest ers, Association of American Ge­ Swetnam, Thomas W. 1983b. Fire Service, Rocky Mountain Forest ographers Program and Ab­ scar dates for Sierra Los Ajos, Son­ and Range Experiment Stn.: 8-14. stracts, AAG Annual Meeting, ora, Mexico. Final Report pre­ Dieterich, John H. 1983. Fire history Phoenix, AZ, April, 6-10 1988: 167. pared for the U. S. Department of of southwestern mixed conifer: A case study. Forest Ecology and 80 Management 6: 13-31. Dieterich, John H.; Hibbert, A. R. 70 1644-45 1988. Fire history in a small pon­ derosa pine stand surrounded by 13 eo 1677-78 chaparral. This volume. s.... 18Q9- Dieterich, John H.; Swetnam, Tho­ ~ 50 1707-08 1 00 CJ 1747 1803-04 mas W. 1984. Dendrochronology rn 1 17B5-BB • 1BB4 of a fire-scarred ponderosa pine. 4-0 1761 1664 rn 1791 1814 Forest Science 30(1):238-247. Q) 1775 ~ 30 1828 Fritts, Harold C. 1986. Historical 1B9 changes in forest response to cli­ E-t .. matic variations and other factors ~ 20 deduced from tree rings. In: Titus, 10 James G., ed., Effects of changes in stratospheric ozone and global cli­ mate, Volume 3: Climate change, lBBO 1900 proceedings of the International Conference· on Health Effects of Ozone Modification and Climate Figure 9.-Severe and very severe El Nino events in relation to percent fire scar-susceptible trees scarred per year from five Gila National Forest chronologies (series 7, 8, 9, 10, and 11 in Change, U.S. Environmental Pro­ fig. 1 and table 1). The means for non-EI Nino and El Nino years were 6.9% and 3.6% respec­ tection Agency: 39-58. tively (p = 0.07, Mann-Whitney test).

16 Agriculture, Forest Service, Rocky Mountain Forest and Range Ex­ periment Station, Tempe, Arizona. Unpublished report on file at Laboratory of Tree-Ring Research, University of Arizona, Tucson. lOp. Swetnam, Thomas W .; Dieterich, John H. 1985. Fire history of pon-. derosa pine forests in the Gila Wil­ derness, New Mexico. In: Lotan, James E.; Kilgore, Bruce M.; Fis­ cher, William M.; Mutch, Robert W., tech. coords. Proceedings - Symposium and workshop on wil­ derness fire; 1983 November; Mis­ soula, MT. Gen. Tech. Rep. INT- 182. Ogden, UT: U.S. Department of Agriculture, Forest Service, Intermountain Forest and Range Experiment Station: 390-397. Swetnam, Thomas W .; Baisan, Chris­ topher H.; Brown, Peter M.; Ca­ prio, Anthony C. 1988. Fire history of Rhyolite Canyon, Chiricahua National Monument. Final Report to U.S. Department of Interior, National Park Service, Southern Arizona Group Office, Phoenix. Contract PX 8601-7-0106. Unpub­ lished report on file at Laboratory of Tree-Ring Research, University of Arizona, Tucson. 36 p. U.S. Department of Agriculture, For­ est Service. National Forest Fire Report. Annual reports 1975-1986. Washington, DC.