Forest IZcolog~ and ?lanagement ELSEVIER

Dendroecological analysis of a cupressoides and a Nothofagus nitida stand in the Cordillera Pelada, Chile

Margaret S. Devall aq * , Bernard R. Parresol b, Juan J. Armesto ’ Southern Research Station. P 0. Box 227, Stone~ille. ,WS 38776. USA h Southern Research Station, P.O. Box 2680. Asherrlle. NC 28502. USA L Laboratorio de Sistemcitica y Ecologia Vegetal. Facrrltad de Ciencias, C’nwersidad de Chile, Casilla. 653. Sontrqo. Chfle Received 6 August 1997; accepted 16 December 1997

Abstract

Lumbering of Fitzroya cupressoides in Chile began in 1599 and continued until 1976, when the was declared a national monument and cutting of live was prohibited. Today, F. cupressoides is threatened; many of the remaining stands in the coastal range appear to be declining, with a predominance of standing dead stems and patchy, sparse regeneration. We performed -ring analysis on a F. cupressoides stand and a nearby Nothofugus nitidu stand, in the Cordillera Pelada. of coastal Chile (40%) in order to examine the ecological history of two stands in the montane forest. Our analysis demonstrates that the F. cupressoides stand has undergone several periods of growth release and disturbance; the last 34 years of the chronology show a trend of increasing growth. In contrast, after 1865 radial growth of the N. nitida stand is fairly constant and steady. Radial growth of these two species is highly correlated with spring rainfall. November rainfall of the current growing season proved to be the best growth predictor of F. cupressoides, whereas current June and December rain, and December rain of the past growing season best predicted growth of N. nitidu. Although episodic disturbances have occurred, the chronologies demonstrated that these stands are vigorous, despite the presence of dead stands nearby. These results do not support the idea that climatic deterioration is responsible for the decline of F. cupressoides during the past 77 years. La tala de F. cupressoides en Chile empez6 en 1599 y continu6 hasta 1976, cuando la especie fue declarada monument0 national y la cotta de arboles vivos fue prohibida. Esta especie es hoy considerada vulnerable. Muchos rodales que subsisten en las cumbres de la cordillera de la costa (40-42”s) parecen estar en declination con una predominancia de individuos muertos en pie. La regeneracidn de F. cupressoides esta limitada a algunos rodales intervenidos. En este estudio se llev6 a cabo un analisis de anillos de crecimiento de F. cupressoides y de N. nitida en dos rodales de la Cordillera Pelada (800-900 m), en la costa de Chile a 4O”S, con el proposito de reconstruir la historia de estos bosques montanos en la cordillera de la costa. El analisis sugiri6 que el rodal de F. cupressoides ha sufrido varios periodos de liberacidn de crecimiento y perturbaci6n. Los hltimos 34 aiios de la cronologia demuestran una tendencia de crecimiento alto. Por otra parte, despues de 1865, el crecimiento de1 rodal de N. nitidu es medianamente constante. El crecimiento de las dos especies se correlaciona fuertemente con las precipitaciones de primavera registradas en una estacidn al sureste de1 area de estudio. La lluvia de1 mes de noviembre de1 aiio anterior fue el mejor predictor de la cronologia de F. cupressoides, mientras que las

* Corresponding author. Tel.: + l-601-686-3161; fax: + 1-601-686-3195.

0378-l 127/98/$19.00 0 1998 Elsevier Science B.V. All rights reserved. PI2 SO378-1 127(98)00221-7 136 Lf s. Del uil Pf id i Fl,WSf EL~oimy iud ,Cf~in<~qrmrnt /ox i iYY8/ iii- i-l>

Ilu\~as de JUII’O 4 diciembre de1 atio actual y Ia Iluvia de diciembre de 10s dos aiios antenores predicen /a cronologia de .V. ~rrr&. X pesar de la presencla de dibturbios ocaslonales. v la exlstencla de rodales Lecinos compuestos de 6rboles muertos y de alta mortandad en rodales maduros. las cronologias demuestran que IOF rodales estudiados de F. crrprrs~o&.~ y de .V. r~itrtin e&n saludables. Estos resultados no apoyan la idea de que el deterioro del clima podtia jer responsable por la reduccl6n de F. clrpressordes durante 10s liltimos 77 adios. :O 1998 Elbevier Science B.V. All rights reserved.

~e\~ordc 4lsrcr: Climate effects: Co~gue: Drndrochronology; Dlsturbancr: Threatened

1. Introduction help assess the relative roles of human impact and climatic change in the current decline of F. cupres- Structural changes that occur in an ecosystem soides. over time are best investigated with frequently moni- F. cupressoides occurs in the Cordillera de la tored permanent plots, but such data are unavailable Costa to the northern part of the island of Chilok, from many forests. Most studies of forest succession and in the Cordillera de 10s Andes between about have been of short duration and provide limited 40“30’ and 43”3O’S (Ramirez and Riveros, 197.5; information about long-term changes. Den- Lara, 1991). Because of the beauty and durability of drochronology can provide a retrospective method of its wood, F. cupressoides was one of the first obtaining knowledge about the history of a forest. species to be commercially exploited in Chile. On Tree growth rings are available for previously un- the island of ChiloC and the area around the estuary studied forests, and tree rings provide a chronicle of of Reloncavi, cutting of F. cupressoides continued the effects of climate on tree growth, and of stand- at a rapid rate from 1599 to the end of the 19th wide disturbances caused by fire, insects or pollu- century (Fonck, 1896, in Donoso et al., 1990). By tion, for example (Cook et al., 1987; Veblen and that time all the accessible forests near Puerto Montt Lorenz, 1987; Van Deusen, 1987, 1989; Veblen et and the estuary of Reloncavi had been nearly elimi- al., 1991, 1992). A ring width chronology, the aver- nated, and the F. cupressoides forests of the central aged ring width measurements from a number of plain between Puerto Montt and Lake Llanquihue trees from a site (F&s, 1976), can provide informa- were cut or burned by colonists in the second half of tion about the history of the forest. The common the 19th century, during the German colonization signal among the sampled trees indicates stand-level, (Armesto et al., 1995). Beginning in 1964, protection year to year influences on radial growth (Cook et al., of the remaining F. cupressoides forests began, but 1987). considerable lumbering continued until 1976, when In this study, we carried out tree-ring analysis of a the species was declared a natural monument, and Fitzroya cupressoides (Mol.) Johnson stand and a cutting was prohibited (Donoso et al., 1990). Norhofugus nitida (Phil.) Krasser stand in the Little is known about the history of disturbance in Cordillera Pelada, the coastal range of Chile at 40% these forests. Regeneration of F. cupressoides is in order to better understand the ecological history of patchy and inadequate in some areas; it seems to be the forest community (Fig. 1). The objective of this limited to areas with low tree cover or large dis- study was to provide a long term historical growth turbed areas, and is extremely poor after clearcutting record of the F. cupressoides and N. nitida stands. followed by fire or grazing (Veblen and Ashton, F. cupressoides, locally known as alerce, is a threat- 1982; Lara, 1991; Donoso et al., 1993; Paez and ened species (CONAF, 1989). The predominance of Armesto, unpublished data). Regeneration is com- standing dead stems, a phenomenon noted over a pletely lacking in old-growth stands on the coastal century ago by Philippi, 1865 and still poorly under- range. In the Cordillera Pelada, disturbed areas ap- stood today, suggests that many F. cupressoides pear to have been created by human-set fires, al- stands in the coastal range are declining, although though natural fires may have also been important in the time of death of trees in many stands is un- the past. Extensive patches of standing, bleached known. Knowledge of the history of this stand may dead F. cupressoides common in many areas of the Fig. 1. Location of the Cordillera Pelada in the southeastern section of the province of Valdivia, Chile. Shaded areas represent desmoyed forest oc bare ground

Cordillera Pelada, side by side with green patches, northward along the Chilean coast and up the river could be the result of such disturbance. Exploitation valleys into the high cordilleras in northern Llanqui- of standing dead stems is still legal, even though the hue (Chudnoff, 1984). Regeneration of N. nitida is logging of dead F. cupressoides may severely im- dependent on large scale disturbances and canopy pair the possibility of natural regeneration because of gaps created primarily by landslides and multiple interference produced by logging debris (Donoso et tree falls krrnesto et al., 1995). Nothofugus wood is al., 1993). commercially valued as an all-purpose timber in Lara and Villalba (1993) suggested that the Chile, but since it is more readily available in low species’ longevity makes it very valuable for den- elevation forests than F. cupressoides, the high ele- droecological studies (long chronologies possible), vation stands have not been subjected to fiie or to and increases the reason to protect it. Boninsegna large scale logging operations. N. nitida has not and Holmes (1985) were the first to successfully been logged or impacted by humans in montane crossdate F. cupressoides and develop a tree-ring coastal forests, in contrast to what happened with F. chronology. Villalba (1990) and Villalba et al. (1990) cupressoides. The wood is of medium hardness and used an F. cupressoides growth chronology from the weight, straight grained and fine textured. Nothofu- Patagonian Andes in Argentina to develop a 1120- gus species are capable of living 200 + years and year reconstruction of summer temperatures. Lara generally have distinct growth rings (Kribs, 1968). and Villalba (1993) used F. cupressoides tree-ring This makes N. nitidu valuable for dendroecological records from southern Chile to estimate summer analysis and for comparison with F. cupressoides temperature for the past 3622 years. forests, that have been more disturbed by humans in N. nitidu is distributed from below 38’S latitude the recent past. Below the summit on a fairly flat area of a southeast facing slope of the cordillera. at around 650 m, was a mixed species stand where a \lgnlf‘i- F. c,rc~)re.r.rnitlr.r samplmg was conducted within cant proportion (15-20%) of the canopy trees were the boundaries of .Clonumento National Alerce Cos- ?I. nitida. This stand had a fairly cloTed canopy, and tero (-IO?O’S. 73’3O’W) a 2308 ha forest preserve a stream running through it. (Fig. 3). located above 800 m elevation in the We cored 30 trees per stand and took two cores Cordillera P&da. the coastal range north of Val- per tree, 180” from each other. In transit. one pack- divia, Chile (39”18’S). The study area receives more age was damaged so we ended up with JO cores than 3000 mm of annual rainfall (Veblen and Ash- from 28 F. cupressoides trees and 42 cores from 21 ton, 1982). with 40% concentrated between June and N. nitida trees. Increment cores were extracted at August (winter). Mean annual temperatures are usu- breast height (1.37 m>. Trees were cored parallel to ally mild, due to the strong oceanic influence, al- the contour. The F. cupressoides stand appeared though local records are lacking. January (mid- fairly homogeneous so we chose sample trees more summer) and July (mid-winter) mean temperatures in or less uniformly spread across the central 5 ha of Rio Bueno (30”18’S; 58 m elevation) are 16.4 and the site. In the mixed species stand we walked a 6.9”C, respectively. High elevations of the coastal more or less straight transect heading northwest into range may receive some precipitation as snow during the stand. About every 15 m we stopped and looked July and August. These coastal forests are far from for a N. nitida tree within about 3 m of the spot. If a industrial sources of air pollution that could cause suitable tree was not found (we bypassed intermedi- forest decline, and receive uncontaminated precipita- ate and suppressed crown class trees, and trees with tion. obvious defects), we went another 15 m and so on. On the summit plateau, (800-1000 m), stands We assumed this gave us a representative cross-sec- dominated by F. cupressoides (Cupressaceae), grow tion of the stand. on heavily podzolized soils (Veblen and Ashton, The cores were air dried, mounted on wooden 1982). These stands have an open canopy 15-20 m blocks, and hand sanded with three grades of sandpa- in height on flat, often poorly drained, sites. The per. Cores were then inspected, with the best 34 shrub layer is formed by Desfontainea spinosa R. cores of F. cupressoides (28 trees) and 33 cores of and Pav., Philesia magellanica Gmel., and extensive N. nitida (24 trees) selected for measurement and patches of the bamboo, Chusquea sp. Below 700 m, crossdating. Cores with indistinct rings or with sev- on steeper slopes and deeper soils, F. cupressoides eral breaks or rotten spots were eliminated (six F. forests are replaced by a mixed forest, with a denser cupressoides and nine N. nitida cores). Ring widths and taller (> 20 m) canopy of N. nitida (Fagaceae), were measured to 0.001 mm using software devel- Drimys winteri Forst. (Winteraceae), and the oped at the Institute for Quantitative Studies (Van Podocarpus nubigenus Lindl. and Saxegothaea con- Deusen, 1993). Reliance on marker rings, interactive spicua Lindl. (both ). Weinmannia tri- graphics, and correlations of ring patterns helped cosperma Cav. (Cunioniaceae) is occasionally pre- establish correct ring sequences, and dating was sent in the mixed forest. assisted and checked by the use of the program Away from the meadow areas on the summit XCHECK (Van Deusen, 1993). Table 1 lists the plateau, at around 900 m, was a nearly pure stand of correlation of each core against the others, and pro- alerce, 20 ha or more in size, suitable for den- vides the crossdated core age, for each species. The drochronological study. The topography was gently oldest F. cupressoides was 398 years and the sloping and the understory was shrubby in places, youngest was 143 years; the oldest and youngest N. but in general fairly open. The forest floor was nitida were 231 and 35 years, respectively. For carpeted in vegetation. Many of the trees had epi- purposes of looking at the long term stand growth phytes growing on the trunks, attesting to the high pattern and for detecting periods of disturbance/re- humidity of the area. We saw no evidence of recent lease, a horizontal straight line standardization is fires in the stand and the trees appeared healthy. recommended by Veblen et al. (1991, 1992). While 1593 0.19 232 1760 1727 0 I9 724 176X 1724 0 75 170 IS22 l7iO 0 71 l-11 1851 1731 0 u I31 IX61 1733 0 74 129 IX63 1’33 0.41 I26 1866 I735 0 ‘3 120 1x72 ! -t _* h 0.19 II8 1874 . -11 0 29 II6 1876 1711 0 24 111 18X1 I 712 0 29 106 I886 1717 0.34 LO4 1888 17-15 0.27 I01 1891 I T-lb 7, u.20 I01 1891 1’7% 0.22 99 1893 ITi2 0 39 92 1900 1751 0.26 92 I900 1752 0 20 91 1901 1751 0.23 89 1903 1755 0.41 89 1903 1757 0.27 81 1911 1758 0.25 79 1913 1’62 0.38 72 1920 1762 9 0.36 69 1923 1765 8 0.28 65 1927 1765 7 0.38 62 1930 1765 13 0.20 60 1932 1774 21 0.51 59 1933 1780 31 0.35 57 1935 1797 3 0.49 38 1954 1n15 32 0.52 38 1954 Id33 I2 0.24 36 1956 1848

reducing [he withm and among core variance, the (1982) postulated that the following components form age-related effect is retained. Let MSRW represent the ring widths of a single tree at time r: mean \tindardized ring width. Percent change in R(t) = C, + B, + Dl, + D2, + e,, radial growrh is computed as MSRW - (1) lagiRlSRW)/tag (MSRW) X 100, where lag means where C, is the climate component at time t, com- an ,~-ner value. A gtowtn release was defined as a mon to all trees; B, is the biological growth trend; 40% or greater change m MSRW sustained for at Dl, is a disturbance signal unique to the individual :e;c% 3 years. We

1821 there is a dramatic surge in growth indicating 2. Results some major disturbance occurred (selective logging, blowdown or fire) which thinned the stand and re- Most of the F. cupressoides trees reached breast leased the surviving trees from competition. Within 3 height between 1730 to 1760 (Table 1 and Fig. 2a). years the radial growth rate increased by 52% and by Only one tree predates 1727. Starting in 1730 initial 1827 the percent change was 86%. A second gradual growth rates were high, as expected for trees coloniz- decline in growth occurred between 1836 and 1896. ing an open site following a large scale disturbance, Another release occurred, though not as dramatic as either logging or fire. A gradual radial growth de- the first release, and by 1899 radial growth rate was cline occurred from 1751 to 1821. This pattern prob- up 44%. After 1908, a sharp decline occurred until ably reflects some ageing effects but most likely 1918. A seemingly cyclical pattern sets up with an represents stand closure (competition effects). After increase to 1934 then a decline until 1945, increasing until 1952 and declining until 1957, with a gradual increasing trend over the next 34 years to 1991. ~ (a) where the chronology ends. The rise tn the chronol- ogy after 1918 represents a third release in the stand. with the growth rate up by 46% within 3 years. Several strong El ?&ho events occurred between 1900 and 199 1 (Diaz and Markgraf, 1992). We examined the F. cupressoides chronology for changes in growth during these years, or the year following, but no clear pattern emerged in the chronology as a result of El Niiio events. .-, ,:_ Tree ages for N. nitida are more variable, as is _ L 2 typical of mixed species uneven-aged stands depen- dent on gap regeneration (Table 1 and Fig. 4a). In 1822 a third tree enters the chronology, then a fourth and fifth tree enter in 185 1 and 1861, respectively. The chronology during this 40 year time period indicates that the oldest trees did not initially grow rapidly. Thus, this stand is not a first generation post disturbance stand like the F. cupressoides stand. Most of the trees enter the chronology after 1880 with little variation in the long term growth trend, in contrast to the F. cupressoides chronology. Only precipitation data from 193 l- 1960 were available from La Union, but November, February and June rain of the current growth season from that location proved to be the best predictors of the F. Fig. 3. (a) The observed (solid) vs. predicted (dashed) value of the chronology. The variation explained in F. cqv-essoides data. The expected values were calculated by the cupressoides program DYNAOLS with the climate variables of current Febru- the transformed chronology was 60% (R2 = 0.60) ary and June rainfall and November ram lagged 1 year. The and the root mean square error (RMSE) was 0.09 observed and predicted values are similar and almost always run ( p < 0.001). The individual correlations between the in the same direction. (b) The smoothed residuals from using the transformed chronology and the climate variables program DYNAOLS with the climate variables current February was - 0.30 for February rain, 0.05 for June rain, and and June rainfall and November rain of the previous year The smoothed residuals will yield a postttve value when the model 0.65 for November rain of the current growth season. underpredicts growth and a negative value when the model over- In contrast, the variation explained by the tempera- predicts. Units on y axes are 0.001 mm. ture and precipitation at Valdivia was only 10% ( R2 = 0.10) and not significant; the Puerto Montt and Punta Galera data gave similar results. Fig. 2b predicted values are similar; the smoothed residuals shows the transformed average chronology of the F. (Fig. 3b) yield a positive value when the model cupressoides ring widths over time. Because we underpredicts ring width and a negative value when used the first difference of the ring widths, the predicted growth is greater than observed. Smooth- transformed chronology shows a poor year following ing involves using information that is available up to years in which growth is above average. The ob- time t as well as information produced after time f served and expected values of the ring widths over (Anderson and Moore, 1979). time are shown in Fig. 3a. The predicted growth was As with F. cupressoides, the La Union weather obtained from the program DYNAOLS with the data correlated the best with the transformed N. climate variables of November, February and June nitidu chronology (Fig. 4b). June and December rain rain of the current growth season. The observed and of the current year and December rain lagged by one of the climate variables can reveal long-term growth trends. November (spring) rainfall ot’ the current growth season proved to predict the F. cupres~ide.~ chronology better than other months of the current or past growth season. This is springtime in the south- em hemisphere, and the current year’s growth begins during the previous year. It is not surprising that spring rain is important in predicting the growth of F. cupressoides. Growth m this month apparently 361 has a significant Influence on the vanation in total l-l 1q annual growth. These results are similar to those Of, reported by Phipps (196la,Phipps, 196lb) who found 1760 1800 1840 1880 1920 1960 2m that (based on weekly dendrometer measurements) Year as much as 80% of radial growth of several decldu- ous species in Ohio occurs during late spring (mid May and late June). --I (b) Although the months selected best predicted ring i t I widths over all the years (November. February and -1 June rain of the current growth season), they may not have been the best for any individual growth season. Rainfall and temperature during other months also influenced growth. December rain of the current and past growth season, and June current season rain predicted the N. nitida chronology best, but N. nitida does not have as strong a climate signal as F. cupressoides as evidenced by its lower R’ (0.46 vs. 0.60) but essentially same RMSE (0.10 vs. 0.09) Fig. 4. (a) Mean horizontal line standardized N. nirida ring-width from model 1. Because bud formation, storage of chronology Illustrating growth and disturbance patterns. Standard- photosynthates, formation of growth hormones and ization was achieved by dividing observed ring widths by the other growth processes occur the year before radial mean ring-width value for each tree. (b) The standardized average growth develops, previous variations in climate can chronology of the N. nitida tree-ring series. affect the ring width of the current year (D’Arrigo and Jacoby, 1992). The horizontal line transformed chronology (Fig. growth season were the best predictor variables. A 2a) demonstrates that this F. cupressoides stand has total of 46% (R2 = 0.46) of the variation in the undergone several periods of growth release and chronology was explained by the precipitation data, disturbance giving the growth trend a roller coaster and the RMSE was 0.10 ( p = 0.002). The individual appearance. The pattern of juvenile growth and grad- correlations between the transformed chronology and ual decrease from 1730 to 1821 probably reflects the climate variables were -0.36 for June rain, 0.29 some ageing effects but most likely represents stand for December rain and 0.37 for December rain of the closure (competition effects). The striking surge in past growth season. growth after 1821 indicates some major disturbance occurred which thinned the stand and released the surviving trees from competition. F. cupressoides 3. Discussion wood persists for well over a century with little decomposition, but there is no evidence of logging Examining the F. cupressoides and N. nitida (stumps) or of blowdown (boles on the ground); ring-width sequences graphically over time in terms most likely, fire caused the releases. The forests of the cordillera are affected by severe west of the cordillera (Veblen et al.. 1976). How- cyclonic storms and were subjected to anthropogenic ever, Heusser (1982) reconstructed average Jnnual fires and cutting (Heusser, 1982) prior to the forma- precipitation and average January temperature at the tion of the Monument0 lUatural Alerce Costero. The town of Alerce (150 km southeast of the study area) last 34 years of the F. cupressoides chronology (31’25’S, 72”54’W). and suggested that temperature \how a trend of increasing growth which coincides there has decreased and precipitation has increased with the exclusion of direct human impact in the since approximately 1920, after an increase m tem- stand (protection of the forests began in 1964). but is perature and decrease m precipitation that continu- not necessarily caused by it. The horizontal line ally worsened during the preceding 350 years. The transformed chronology of N. nitidu (Fig. 4a), from decline in temperature in Puerto IMontt since the mid-elevation in the cordillera. indicates some type early 1900’s has been documented by Aceituno et al. of disturbance around 1865. After that there is a (1993). Our analysis demonstrated that growth of F. strong recovery and a more-or-less flat long term cupressoides in the stand we studied (at 900 m) was growth trend, which (as opposed to the short term weakly correlated with temperature and precipitation year to year sawtooth variation typical of all tree at Valdivia (sea level). and highly correlated with chronologies) is fairly constant and steady, in con- precipitation at La Union (approximately 30 km trast to the cycling observed in the F. cupressoides southeast of the study area>. chronology. The growth trend indicates a healthy Villagran (1991) reports that F. cupressoides and stand undergoing natural processes in uneven-aged N. nitidu forests were present at mid elevations on mixed species stands. Although the chronologies re- the coastal range between 12,500 and 9500 YBP veal that these stands of F. cupressoides and N. (from the beginning of the Holocene). Hence, de- nitidu are generally healthy, in nearby stands some cline of F. cupressoides pollen in profiles from the or all of the F. cupressoides trees have died stand- Lake district may indicate that the species was mi- ing. grating to higher elevations, both in the Andes and Heusser (1982) identified fossil pollen in three coastal ranges rather than declining regionally. Pollen cushion bogs of the Cordillera Pelada. He found F. profiles show that F. cupressoides has colonized the cupressoides type pollen in the Cordillera Pelada Andes in the past 3000 years (Villagran, 1991). diagrams after about 6500 years Before Present (BP), It is surprising that the F. cupressoides stand has when a regional rise in precipitation took place been healthy, with no long term decline, while dead (precipitation reconstructed from fossil pollen data). stands and stands of mixed dead and live trees occur Although fire and exploitation by European colonists nearby in the Cordillera Pelada and on Chiloe Island have accentuated the species’ decreasing numbers in (Armesto et al., 1995). In some of the dead stands, recent centuries (Veblen et al., 19761, evidence from there is evidence of recent fire, but in others it is not three pollen records at the town of Alerce demon- obvious why the trees have died. There are no strated that the greatest proportion of F. cupres- obvious symptoms of insects or disease, and no soides pollen was present during the Holocene around nearby sources of pollution. It appears that weather 4000 years BP, and that F. cupressoides pollen has conditions at Cordillera Pelada are more similar to gradually declined since then. Heusser and Streeter conditions to the southeast than to those at Valdivia. (1980) postulated that the decline began before the The long decrease in precipitation at the town of arrival of the Europeans and likely was caused by Alerce began to reverse 77 years ago; perhaps some the estimated reduction in precipitation. Heusser of the trees that died are exhibiting a delayed reac- (1982) also suggested that past rainfall in the tion to the previous unfavorable conditions. Because Cordillera Pelada was distributed more uniformly dendrochronology does not show any evidence for a through the year, with increased humidity and decline in growth rates associated with decreased cloudiness and lower temperature, and without a precipitation, some other factors must be responsible summer drought. A decrease in precipitation since for the observed dying stands. Delayed reaction to the middle of the 19th century was documented by previously unfavorable conditions could be a plausi- instrumental records at Valdivia about 40 km north- ble alternative explanation, but climatic deterioration 144 .M S. Dcmll et ui. / Forest Ecology urld Mmugement 10X (19981 135-l-15

IS not supported by the analysis of N. nltidn in a ICE ot‘ Ficrow crcprrssurdes forests m bouthem Chde J nearby stand. Vepetatlon SC!. 4. 303-312. Fonck. F.. 1896. ViaJes de Fray Francisco MenCndez a la corddlera (Travels of Fnar Francisco Menendez to the mountams). Nlemeyer. Valparaiso. Chde. Acknowledgements Fntts. H.C., 1976. Tree Rmgs and Chmate. Academic Press. London, England. Grayblll. D.A., 1982. Chronology development and analysis. In: The U.S. Office of International Cooperation and Hughes, M.K.. Kelly, PM.. Pdcher. J.R.. Lamarche, U.C . Jr. Development provided funding for the study. J.J. (Eds.). Climate from Tree Rmgs. Cambndge Univ. Press. Armesto acknowledges the support of Fondecyt 92- Cambndge. England. 1135. We thank Kim LG and Sukanya Koppolu for Heusser, C.J.. 1982. Palynology of custuon bogs of the Corddlera measuring and crossdating the cores. We thank Tom Pelada Provmce of Valdivla, Chile. Quaternary Res. 17, 7 l-92. Heusser, C.J.. Streeter, S.S.. 1980. A temperature and preclplta- Doyle and Michael Cane for reviewing the tion record of the past 16.000 years in Southern Chde. Science manuscript. 210, 1345-1347. Kribs, D.A., 1968. Commercial Foreign Woods on the American Market. Dover Publications, New York. Lara, A., 1991. The dynamics and disturbance regimes of forests References in the south-central Andes of Chile. PhD Dissertation, Geogra- phy Department, University of Colorado, Boulder, CO. Aceituno. P.. Fuenzalida, H., Rosenblath, B.. 1993. Climate along Lara, A., Villalba. R., 1993. A 3620-year temperature record from the extratropical west coast of South America. In: Mooney, Fifzroyn cupressoides tree rings in southern South America. H.A., Fuentes. E.R., Kronberg. B.I. (Eds.), Earth System Science 260. 1104- 1106. Responses to Global Change. Academic Press, San Diego, Philippi, R.A., 1865. Excursidn bottica, en Valdivm desde Los CA, pp. 61-69. Cuncos en el departamento de La Uni6n atraves de la Cordlllera Anderson, B.D.O.. Moore, J.B., 1979. Optimal Filtering. de la Costa. [Botanical excursion in Valdivia from Los Cuncos Prentice-Hall, Englewood Cliffs, NJ. in the department of La Uni6n across the Coastal Range] Ann. Armesto, J.J., Villagrti, C.. Aravena, J.C., Ptrez, C.. Smith- Univ. Chile 27, 289-351. Ramirez, C., Cortis, M., Hedin, L., 1995. forests of Phipps, R.L., 1961a. A preliminary report concerning radial growth the Chilean coastal range. In: Enright, N., Hill, R. (Eds.). of selected trees at Neotoma. Ohio Agncultural Station Circu- Ecology of Southern Conifers. Melbourne Univ. Press, Mel- lar, 102. bourne, Australia. pp. 156-160. Phipps, R.L., 1961b. Analysis of five years’ dendrometer data Boninsegna, J.A., Holmes, R.L., 1985. Firzroya cupressoides obtained witbin three deciduous forest communities of yields 1534-year long South American chronology. Tree-Ring Neotoma. Ohio Agricultural Experiment Station Circular, 105. Bull. 45, 37-42. Ramirez. C., Riveros. M., 1975. Los alerzales de la Cordillera Chudnoff, M., 1984. Tropical Timbers of the World. Agriculture Pelada-flora y fitosociologia. [The alerce forests of the Handbook 607. U.S. Forest Service, Forest Products Labora- Cordillera Pelada-flora and phytosociology] Medioambiente 1, tory, Madison, WI. 3-13. CONAF. 1989. Red List of Chilean Terrestrial Flora. Corporaci6n Van Deusen, P.C.. 1987. Testing for stand dynamics on red spruce National Forestal, Santiago, Chile. growth trends. Can. J. For. Res. 17. 1487-1495. Cook, E.R., Johnson, A.H., Biasing, T.J.. 1987. Forest decline: Van Deusen, P.C., 1988. Red spruce tree ring analysis using a modeling the effect of climate in tree rings. Tree Physiol. 3, Kaltnan filter. In: Van Deusen, P.C. (Ed.), Analyses of Great 27-40. Smoky Mountain Red Spruce Tree Ring Data. General Tech- D’Artigo. R.D., Jacoby, G.C., 1992. A tree-ring reconstruction of nical Report SO-69. U.S. Forest Service, Southern Forest New Mexico winter precipitation and its relation to El Experiment Station, New Orleans, LA, pp. 57-67. Niiio/Southem Oscillation events. In: Diaz. H.F.. Markgraf, Van Deusen, P.C., 1989. A model based approach to tree ring V. (Eds.), El Niiio: Historical and Paleoclimatic Aspects of the analysis. Biometrics 45, 763-799. Southern Oscillation. Cambridge Univ. Press, Cambridge, Van Deusen, P., 1993. Dynaclim Version 3.2 Users manual. U.S. Great Britain, pp. 7-28. Forest Service, Southern Forest Experiment Station, Institute Diaz. H.F., Markgraf, V., 1992. El Niiio: Historical and Paleocli- for Quantitative Studies, New Orleans, LA. matic Aspects of the Southern Oscillation. Cambridge Univ. Veblen, T.T.. Ashton, D.H.. 1982. The regeneration status of Press, Cambridge, Great Britain. Fitzroya cupressoides in the Cordillera Pelada, Chile. Biol. Donoso, C., Grez, R.. Sandoval. V., 1990. Caracterizacibn de1 tip0 Conserv. 23, 141-161. forestal alerce. [Characterization of the alerce forest type] Veblen, T.T., Lorenz, D.C., 1987. Post-fire stand development of Bosque 11, 21-34. Ausrrocedrus- Norhofagus forests in northern Patagonia. Veg- Donoso, C., Sandoval, V., Grez, R., Rodn’guez, J., 1993. Dynam- etatio 71, 113-126. .M S. Drtull et ~1. / Forett Ecologv und .Munu,~emmr ION f lYYRi /_iS- 145 I.Li

Veblen. T Delmahtro. R.. Schlatter. J 1976 The conservation of Chtle durante el Tardlglaclal y Postglacial. [HIatop of the F ircprrrsotd~s and It5 enrtronment m Southern Chile. Enw remperate forests of south Chrle dunng the late gluc~al and ran. Conserv 3. 291-301. postglaclal] Rev. Chllena Htstoria Nat. M, U7-460 Vehlen, T.T. Hadley. K.S.. Reid, M.S.. Rebertus. A.J.. 1991. Vlllalba, R., 1990. Climattc tluctuatlons in northern Patagontd \lethods of dstectmg past qx~~e beetle outbreaks in Rocky dunng the last 1000 years as Inferred from tree-rmg records. llountun sub-alptne forests. Can. _I. For. Res. 21. 212-251. Quaternary Res. 33, 316-360. Veblen. TT., Kitzberger. T.. Lara, A.. 1992. Disturbance and Villaiba. R., Leiva. J.C.. Rubulls. S.. Suarez. J.. Lenzano. L.. forest dynamics along a transect from Andean ram forest to 1990. Climate, tree-nng, dnd glacial fluctuations In the RIO Paragonian shrubland. J. Vegetation Scl. 3. 507-520. Frias Valley, RIO Negro, Argentma. Xrctlc Alpme Req. 72. Villagran. C.. 199 i. Histona de 10s bosques templados del sur de 215-232.