Dendrochronology and Lichenometry: Colonization, Growth Rates and Dating of Geomorphological Events on the East Side of the North Patagonian Icefield, Chile

Geomorphology 34Ž. 2000 181±194 www.elsevier.nlrlocatergeomorph

Dendrochronology and lichenometry: colonization, growth rates and dating of geomorphological events on the east side of the North Patagonian Icefield, Chile

Vanessa Winchester a,), Stephan Harrison b a School of Geography, UniÕersity of Oxford, Oxford OX1 3TB, UK b Centre for Quaternary Science, CoÕentry UniÕersity, CoÕentry CV1 5FB, UK

Received 26 May 1999; received in revised form 13 December 1999; accepted 16 December 1999

Abstract

This paper highlights the importance for dating accuracy of initial studies of delay before colonization for both trees and lichens and tree age below core height, particularly in recently deglaciated terrain where colonization and growth rates may vary widely due to differences in micro-environment. It demonstrates, for the first time, how dendrochronology and lichenometry can be used together in an assessment of each other's colonization and growth rates, and then cross-correlated to provide a supportive dating framework. The method described for estimating tree age below core height is also new. The results show that on the east side of the North Patagonian Icefield in the Arco and Colonia valleys, Nothofagus age below a core height of 112 cm can vary from 5 to 41 years and delay before colonization may range from a maximum of 22 years near water to a minimum of 93 years on the exposed flanks of the Arenales and Colonia Glaciers. Tree age plus colonization delay supplied a maximum growth rate of 4.7 mmryear for the lichen Placopsis perrugosa and lichen colonization is estimated to take from 2.5 to approximately 13 years. A minimum lichenometric date of 1883 was estimated for an ice-formed trimline at the junction of the Arenales and Colonia glaciers and a maximum dendrochronological date of 1881 for a water-formed trimline in the Arco valley. Tree and lichen ages around the valley suggest that a glacial outburst drained the 1881 high level lake releasing approximately 265 million cubic metres of water. Repeated flooding, with a minimum of 38 high lake levels, is suggested by horizontal sediment lines on the Arco valley walls and moraine flanks. Dating confirmed diminishing flood levels with a last minor flood in 1963. The wider significance of the work is that it should produce more accurate dating of recent glacier fluctuations around the North Patagonia Icefield, an area where dated reference surfaces are extremely scarce. q 2000 Elsevier Science B.V. All rights reserved.

Keywords: dendrochronology; lichenometry; colonization; growth rate; trimline; dating

) Corresponding author. Fax: q44-1865-559560. E-mail address: [email protected]Ž. V. Winchester .

1. Introduction the shrinking margin of the same glacier and Villalba et al.Ž. 1990 found a similar optimum rate for the This paper describes the little-explored valleys of species below the Frias Glacier on the northeastern the Arenales, Colonia, and Arco Glaciers on the flank of Mt. Tronador, 600 km north of San Rafael, eastern side on the North Patagonian Icefield, south- with colonization varying from 5±10 years in shel- XX ern Chile, 47816 S, 73813 WŽ. Fig. 1 . It shows how tered spots to 67 years on the exposed valley bottom. dendrochronology and lichenometry can be used to- MercerŽ. 1970 estimated a 70-year delay for gether to date environmental changes and illuminate Nothofagus when working on the western flank of the scale of events following deglaciation. The work the Southern Icefield, while SwedaŽ. 1987 , studying initially focuses on an investigation of the effects of tree age around the Soler Glacier on the eastern side environmental variations on tree age below core of the Northern Icefield, found colonization delays of height, colonization delays, and lichen growth rates. 24 to 30 years on the northern valley side, but 50 This type of investigation, often neglected in previ- years on the valley bottom where there had been a ous geomorphic studies, is mandatory; without it, fire. The longest colonization delay was observed by dating with dendrochronology and lichenometry Nichols and MillerŽ. 1951 who found no trees on the could be seriously misleading. 80-year-old `Little Ice Age' moraines at the foot of Dendrochronological surface exposure dates are the Ameghino Glacier on the eastern side of the derived from tree age, with age calculated by adding Southern Icefield. the annular ring counts taken from cores to an The wide divergence in tree colonization rates estimate of age below core height and years before means that any generalized application of den- colonization. Accurate calculation of all these param- drochronology in this region is inappropriate and, as eters is particularly necessary in areas of recently Warren and SugdenŽ. 1993 emphasized, there is a deglaciated terrain where growth can be slowed by need for detailed case studies. The aim of the present cold katabatic winds, shallow soil on moraine crests, study is to address this need with respect to both extremes of heat and cold, drought and flooding, or dendrochronology and lichenometry which latter late-lying snow on valley bottoms. Following technique relies on accurate estimates of lichen deglaciation, soils are initially sterile and unstable, growth rates. This paper shows how tree age below and potential seedbeds may be distantŽ Matthews, core height and colonization delay and lichen growth 1992. . rate and colonization delay can be investigated using Around the North and South Patagonian Icefields, a minimal dating framework to supply an outline there have been no previous studies of the number of control for the local variables affecting growth. It years of tree growth below coring height and tree concludes by using tree and lichen dating to show colonization rates have only been examined in a the scale and sequence of events in the valleys. single study beside the San Rafael Glacier on the northwestern side of the Northern IcefieldŽ Winches- ter and Harrison, 1994. . There have been, however, 1.1. PreÕious exploration several studies of colonization showing wide vari- ability. HeusserŽ. 1964 believed that Nothofagus Three previous expeditions in 1958, 1963 and seedlings take about 25 years to become established 1979, have explored the Colonia and Arco valleys. near the San Rafael Glacier. Alternatively, Winch- The first and only one to visit the Arco valley was ester and HarrisonŽ. 1994 used Warren's Ž. 1993 the ExpedicionÂÂ Chileno Japonesa Andes Patagonicos observations to estimate a minimum 6-year delay lead by TanakaŽ. 1980 . Tanaka reported local settlers before Nothofagus colonized bedrock cracks above as saying that between 1920 and 1958, there were

Fig. 1. Map of the field area with surface exposure dates provided, in this study, by lichenometry and dendrochronology and glacier ice front positionsŽ. numbered and dashed . V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194 183 184 V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194 floods in the Rio Colonia valley every yearŽ except 2. Study area in 1954. between the end of December and January. These lasted around three days, with water rising to a In the Arco valley, the most important features for maximum height of 7 m above the normal summer this study are the height and position of the various level. Icebergs were, apparently, a common summer moraines and a horizontal trimline marking a change feature before 1928, but after this date, they were in vegetation type and densityŽ. Fig. 1 . There are only seen during floods. An outburst flood had oc- three main moraines in the valley: a 25 m high curred just before Tanaka arrived in 1958 and he recessional moraine lies across the glacier foreland, a photographed stranded icebergs halfway up the distal 90 m high terminal moraine defines the eastern end flank of the terminal moraine at the foot of the Arco of Lago Arco, and on the lake's southern shore is a valley. The 1963 expedition, led by Shipton, de- 40 m high lateral moraine bordering a fluvioglacial scended the Colonia Glacier having crossed the ice- terrace. The trimline, lying approximately 120 m field from the northwestŽ. Shipton, 1964 . The pur- above the lake's surface and 30 m above the crest of pose of the 1979 expedition was to study the geology the terminal moraine, runs along both sides of the of the region, with the party reaching the valley from lower valley as far as two rock spurs projecting from Rio Nef and traveling down the east side of the the valley wallsŽ. Fig. 2 . Immediately east of the Colonia valleyŽ Yoshida, 1981, and personal com- northern spur, a truncated debris cone is cut by the munication. . line, with secondary forest growing below and ma-

Fig. 2. View looking north across Lago Arco showing:Ž. 1 trimline, Ž. 2 sediment lines marking high lake levels on the terminal moraine, Ž. 3 debris flow,Ž. 4 rock spur, Ž. 5 Arco Glacier. V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194 185

Fig. 3. The east face of Cerro Colonia bluffŽ. 1 showing rapid southward decline in the height of deposits left by the Colonia Glacier Ice during the `Little Ice Age' and a residual moraine fragmentŽ. 2 on the outwash plain. ture forest growing above it. The mature forest is Cerro NegroŽ. also named by us river outlet. Recent characterized by large trees with an extremely dense, ablation and push moraines containing large masses shrubby understorey growing up between fallen of dead ice lie between the latter two calving fronts trunks; the secondary forest contains even-sized, filling the centre of the outwash plain. Lateral closely spaced trees with almost no under storey or moraines line the Colonia valley walls and residual fallen trunks. end moraine fragments, eroded by wind and water, Between the 1996 lake level and the trimline, stand on the valley floorŽ. see Fig. 3 . Former out- there is a horizontal sequence of approximately 38 wash streams have created four terraces, marking sand and coarser sediments lining both valley sides different periods of downcuttingŽ Harrison and and all lateral and terminal moraine surfaces free of Winchester, 2000. . Currently, a single river drains vegetation or debris slides. the glacier; it flows some 3 km to join the 11 km At the junction of the Arenales and Colonia long Lago Colonia, which is bounded on its southern Glaciers, lateral moraines are deposited against the shore by a crescent of 65 m high moraines near the valley side below ice scoured bedrock. In 1996, the head of the Rio Colonia valley. Colonia Glacier ended against a rock bluffŽ named by us `Cerro bluff', see Fig. 1. barely closing the 3. Methods Arco valley. The glacier calved into three proglacial lakes: one enclosed at the foot of the Arco valley; A map of the region, produced from a combina- one on its southwestern side below the Cerro bluff; tion of 1979 and 1983 vertical aerial photographs and the other on its northeastern side fronting the and the 1975 1:50,000 map, was field checked on the 186 V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194

Fig. 4. Oblique aerial photograph taken in 1944 looking southeast down the Arco Glacier showing icebergs floating in a greatly enlarged Lago Arco, with the top of the terminal moraine just showing above the water and, on the Lago's south side, the fluvioglacial terrace covered. Note the two rivers cross the Colonia outwash plain in the backgroundŽ photograph supplied by `Instituto Geografico Militar de Chile', Santiago. . ground. Landform profiles were constructed using an on terraces and valley sides up to and within the Abney level and tapes and relative altitude measure- forest margins and on transects from bottom to top ments, accurate to "5 m, were taken of prominent of all glaciogenic features. Cores from 124 trees, landscape features. A 1944 oblique aerial photo- taken with a 5 mm increment borer, were extracted graph, showing a greatly expanded Lago Arco, with from between 6 and 200 cm above ground depending only the crest of the terminal moraine visible above on the gradient and local growing conditions. Dates the surface, provided evidence of a former high lake for surface exposure were derived from ring counts levelŽ. Fig. 4 . plus estimations of colonization delay and age below core height. The cores were extracted from Nothofa- 3.1. Dendrochronology gus speciesŽ including N. pumilio, N. betuloides, N. Below, the trimline trees are scarce; sampling was nitida and N. antarctica. and a record was kept of carried out wherever possible on a selection of trees the location of each tree selected, the core height V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194 187

Fig. 5. Size±frequency histograms of lichen populations in the Arco and Colonia valleys.Ž. a The oldest populations show disturbance with high mortality and episodic recolonization.Ž. b Relatively young populations, mortality has reduced size±frequencies but the distributions are still almost intact.Ž. c Young populations on the Arenales±Colonia moraine, with lichen age decreasing and colonization increasing from ridge top to base, suggesting slow glacier downwasting.Ž. d,e The two youngest populations, with colonization in Ž. d beginning to decline, but still unaffected inŽ. e where it is increasing exponentially. 188 V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194 above ground, tree height in relation to surrounding establishment of lichen growth rates from sizerage trees, the number of touching tree crowns, distance correlations of largest lichens on surfaces of known from other trunks, trunk circumference, slope aspect exposure date; it provides minimum dates for rock and angle; notes were also made of any other factors surface exposureŽ. Beschel, 1950, 1961 . The second affecting growth. The cores were mounted on wooden approach, based on size-frequency histograms and battens and polished; a hand lens was used to count devised by WinchesterŽ. 1989 , is useful for identify- the rings. ing anomalous lichens resulting from amalgamation To establish approximate sizerage relationships of thalli or disturbances caused by movement or below coring heights, 12 small trees were cut at flooding. Disturbances are likely where distributions ground level. Notes were made of each tree's local display bi- or multi-modal frequencies. For the his- environment, its height and its basal ring count. A tograms, a class size of 14 mm was selected by trial record was also made of those trees with sections and error: smaller size increments produced graphs showing narrow rings, compression wood or trauma dominated by noise and at larger sizes, the detail was tissue suggesting stressful growing conditions lost. Populations of less than 35 individuals were Ž.Schweingruber, 1988 . Estimates of age below core rejected as being too small to provide meaningful height were based on the characteristics of the rings information; scarcity of lichen growth in the area nearest the central pith and the recorded details for meant that populations seldom exceeded 90 individu- each tree. Thus, when establishing germination dates alsŽ. Fig. 5 . Where populations are unimodal, size± where multiple stress factors were present, a maxi- frequency distributions can be used for relative dat- mum number of years were added to the ring count. ing by comparing distribution curves. Very young Alternatively, if trees were unstressed and there were populations produce distributions showing exponen- wide rings at the pith end of a core, rapid growth to tial curves indicative of increasing rates of coloniza- core height was assumed and a minimum number of tion, while older distributions show a skewed curve years were added. due to declining colonization as growing space be- comes limited. Colonization ceases with increasing age and average population sizes increase while mor- 3.2. Lichenometry tality begins to reduce size frequencies and gaps appear in the distribution. Approximately 2800 lichen measurements were made, with sampling carried out across selected boulders on terraces and recently exposed valley 4. Results sides and on transects from bottom to top of all lichen-carrying glaciogenic features. The boulders Before any dating could be accomplished, tree selected were those carrying the largest number of age below core height and colonization delay and healthy, approximately circular lichen thalli, and lichen growth rates and colonization were investi- where practicable, all thalli were measured on the gated. The findings are described below, followed by surface carrying the largest population. Notes were estimated dating for surfaces exposed by ice or lake made on location and boulder size, surface aspect level changes. Dates are based on age of the oldest and angle for each population for the purpose of trees and largest lichen measurements in the samples comparison and in an attempt to provide a represen- and include estimated growth to coring height for tative cross-section of the populations in each loca- trees and both tree and lichen colonization delays tion. Measurements were taken of the longest axis of Ž.Table 1 . the species Placopsis perrugosa using a flexible scale accurate to the nearest millimetre; dates for 4.1. Tree age below core height rock surface exposure were based on sizerage correlations with the addition of a colonization delay. Table 2 shows variations in growth rates and tree Two lichenometric approaches were used for the height, with average annual growth rates for young study. The first approach is based on the indirect Nothofagus below a height of 112 cm varying from V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194 189

Table 1 Key tree and lichen dating estimates for lake levels and outburst floods in the Arco and Colonia valley and ice recession at the Arenales and Colonia Glacier junction and the Colonia terminus Sites Tree age Date Lichen sizeŽ. mm Date Arco Valley foot moraine deposits 7 1963 58, 145 1981, 1963 a Valley foot between streams 9q2 1963 Fluvioglacial lateralŽ. 5 mba 16 1958 b Fluvioglacial terraceŽ. 40 m 22q4 1944 450, 456, 476 1898, 1896, 1892 Foreland recessional moraineŽ. 20 mb 14 1956 130 1966

Foreland mid-beach 11q13 1946 250 1940 ba Terminal moraineŽ. 5, 45 m 15 , 29q5 1959, 1936 339, 440 1921Ž. 1937 , 1900 ba c and distal sideŽ. 5, 40, 90 m 12q3 ,16q3 , 31q10 1959-51-29 530 1881 c Debris cone below trimline 82q7 1881 ) ) Valley wall above trimline 116q5 1848

Colonia Cerro Negro crest of recent moraines 62 1980Ž. 1983 Central moraine footŽ. near water level 64 1980Ž. 1983 Central moraineŽ. near water level 64, 113 1980, 1969Ž. 1975 East valley sideŽ. near water level 60 1981Ž. 1983 West valley sideŽ. near water level 53 1982Ž. 1984 Cerro bluff, east sideŽ. 3 mba 10 1964 50 Ž 1.5 m b . 1983 Ž. 1985 b Cerro bluff, east sideŽ. 27 m 23q3 1944 Cerro Negro above crest of recent moraine 321 1925 ba West valley side dry channel bedŽ. 12 m 13q4 1953 Ž. 1957 b West valley side end moraineŽ. 20 m 3q10 1917 b b East valley side terraceŽ. 27 m 49q7 1914 360 Ž. 20 m 1917 Residual moraine on valley floor 420 1904 ArenalesrColonia top lateral moraine base and crest 62dd , 135 1970, 1954, r ecdc Arenales Colonia trimline 17q3 1883 472 1883 b Lago Colonia south endŽ. 65 m 68q7 1897 458 Ž foreshore . 1896

Surface exposure dates, unless otherwise marked, for trees are calculated by adding a colonization delay of 26 years to the ring count, plus growth to core heightŽ given as subscript; where this is lacking, the tree has been sectioned at its base. . For lichens, dates are calculated by dividing the longest axial measurement by a growth rate of 4.7 mm and adding a 2.5 year-colonization delay. Growth rates near water level could be increased to 6 mmryear producing younger datesŽ. bracketed below . a Tree colonization maximum 22 years. b Metres above water level in February 1996. cAutocorrelation. d Lichen colonization minimum of 13 years. e Tree colonization minimum of 93 years.

2.9 to 22.4 cmryear. These rates imply that tree age 4.2. Tree colonization could potentially vary from 5 to 41 years below cores taken at the standard coring height of 112 cm. Three different tree colonization delay periods However, in practice, highly stressed trees growing were deduced from a number of interrelated observa- at the slowest rate are often difficult if not impossi- tions. The trimline and the horizontal sediment lines ble to sample, with multiple branches developing at on the valley walls and moraines in the Arco valley the base of the trunk. In this study, our estimates of display a sequence of at least 39 former high lake tree age below core height were limited to the 15-year levels. The glacier surface may or may not have range suggested by the sectioned trees, including reached trimline height during the `Little Ice Age', considerations of the height of the core above ground, but tree age above the line shows that it did not the tree's characteristics, and local environment. exceed it. TanakaŽ. 1980, Fig. 61 believes that, 190 V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194

Table 2 spectively. The maximum delay for these trees close Varying height±age relationships and average growth rates of to water level is, therefore, 22 yearsŽ 1996y16s Nothofagus speciesŽ including N. pumilio, N. betuloides, N. y s . nitida and N. antarctica. based on ring counts from tree trunks 1980 1958 22 . sectioned at ground level A minimum 93-year colonization delay is esti- Tree heightŽ. cm Ring count Ž age . Growth rate Ž cmryear . mated for trees at the junction of the Arenales and Colonia Glaciers where the valley side, at 450 m 18 3 6 45, 51, 58, 112 5 9, 10.2, 11.6, 22.4 a.s.l., is exposed to winds and violent westerly storms 40, 58 14 2.9, 4.1 sweeping down from the icefield above. The ex- 85 15 5.7 tended delay is based on a lichen date of 1883 for an 126, 130 7 18, 18.6 ice-formed trimlineŽ. see below and the maximum 170 12 14.2 tree age of 20 years for the trees above itŽ 1996y20 177 10 17.7 s1976y1883s93. . Average growth rates are included to show growth range and In the Arco valley, a ring-count of 82 for the provide a measure of comparability between treesŽ. or cores of oldest tree in the secondary forest on the debris cone different heights. below the vegetation trimline added to a 26-year colonization delay and 7-year growth to core height provides an age for the trimline of 115 years and a before the outburst flood, the lake filled until it just maximum estimated drainage date for the highest covered the Colonia terminus; it is likely also to lake level of AD 1881. have covered the Arco terminus. We assumed that the icy, sediment laden waters of the highest lake removed any former trees and 4.3. Lichen growth rate and colonization delay lichens, and we also assumed that once the water retreated, the mostly sterile lake deposits produced If all lichens were removed by the high lake level similar colonization delays and germination condi- before 1881, then a lichen growth rate can be calcu- tions on all stable surfaces except for those close to lated from the sizeŽ. 530 mm of the largest P. the long-term lake level where moisture levels are perrugosa Žgrowing on the top of the terminal raised. A maximum 26-year colonization delay was moraine. divided by the 115-year exposure date for deduced from the difference between the 1944 date the debris cone below the trimlineŽ. Table 1 . A of the oblique aerial photograph showing Lago Arco lichen sizerage correlation of 530r115 is equal to a covering the fluvioglacial terrace and the 1970 ger- growth rate of 4.6 mmryear just below the 4.7 mination date of the oldest tree on the terrace. Re- mmryear rate for Placopsis patagonica at Glacier stricted tree age on the terrace strongly suggested San Rafael on the west side of the icefieldŽ Winches- that any former trees were killed when it was last ter and Harrison, 1994. . The addition of some lesser under water. delay, based on reduced tree colonization on the However, 26 years is a maximum delay, periodic fluvioglacial terrace, would produce an increased high lake levels after 1944 may also have covered lichen growth rate and, except near water, this seems the terrace, as for example, the 1958 flood reported unlikely due to the xeric nature of the Arco environ- by TanakaŽ. 1980 . At midpoint on the Arco foreland ment and the genus' preference for growth in damp beach, ca. 38 m above lake level, there is a tree conditionsŽ. Wirth, 1987 . dating to 1944 showing that the 1958 flood could not A small upward adjustment from 4.6 to 4.7 have covered the fluvioglacial terrace although mmryear was made to allow for a colonization Tanaka's evidence suggests that it is likely to have delay suggested by the absence of lichens on the risen close to the top of the flanking lateral moraine. surfaces exposed at some time after 1980 around the If this is so, then the 26-year estimate for coloniza- terminus of the Colonia Glacier. The delay's length tion delay on the terrace could not apply to 15- and was deduced as follows: if the 4.7 mmryear rate is a 16-year-old trees now growing 5 m above lake level maximum rate not to be exceeded, then the maxi- on the terminal moraine and the terrace lateral, re- mum delay here is 2.5 yearsŽ 530r4.7q2.5s115 V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194 191 years. , with both the delay and the growth rate thus 4.4. EÕents in the Arco Õalley being identical to those at San Rafael. However, on the east side of the icefield at the When the Arco Glacier started to recede from the Arenales±Colonia junction and about 100 m above terminal moraine at the end of the `Little Ice Age', the present ice level, the colonization delay is likely the depth of the lake created by blocked outlet to be longer than 2.5 years for the following reasons: channels under the Colonia Glacier, will have been only the highest of four lateral moraines extending controlled by the thickness of the Colonia ice sur- along the east-facing valley wall carries lichen popu- face. Maximum tree age near the trimline shows that lations. The largest lichen growing at the foot of this the ice front had retreated at least to Lago Arco's moraine in 1996 measured 62 mm; this size repre- midpoint west of the debris cone before the glacial sents a date of 1983. Since there were no lichens outburst in 1881Ž. Fig. 6 . A rough estimate of the below this, the delay before colonization is 1996y former lake below the trimline suggests that the 1983 suggesting a minimum delay of 13 years rather outburst could have released a minimum of 265 than the 2.5 years estimated for the Arco valley million cubic metres of water into Lago Colonia. below. Despite an increased colonization delay, it is Widely differing tree and lichen ages on the assumed that the 4.7 mmryear growth rate remains fluvioglacial terrace, with lichen sizes representing unchanged based on the observed tolerance of P. dating in the 1890s and trees postdating 1970Ž Table patagonica to changes in exposure and temperature 1. , suggest that high water levels were only main- within the local range, as illustrated by its growth in tained for brief intervals of a few weeks before exposed and protected positions beside the San Rafael drainage occurred. The lichen ages infer that the GlacierŽ. Winchester and Harrison, 1994 . lichens survived flooding while the trees did not.

Fig. 6. Diagrammatic interpretation of the relative heights and positions of the moraines, fluvioglacial terrace, Colonia and Arco Glaciers, and high lake levels in the Arco valley in 1881, 1944, 1958, 1963, and 1996. 192 V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194

Lichen survival underwater is supported by research butionŽ. Fig. 5c supplying a minimum rock exposure showing that some crustose lichensŽ e.g., Rhizocar- date of 1970. During the `Little Ice Age', the termi- pon geographicum. can endure submersion at 9±128C nus of the Colonia Glacier rose to a high point for 3±5 weeksŽ. Ried, 1960 and P. perrugosa, with against the Cerro Colonia bluff and extended down its preference for wet conditions, may be able to the valley for around 1.5 km. Its extent is shown in survive longer. After each event, the lake would the 1979 and 1983 aerial photographs and ice contact have drained nearly to its present level controlled by with the valley wall is indicated in Fig. 3 showing the height of the rock step or bar at the southern end the southward sloping, ice-scraped bedrock, and un- of the terminal moraine. vegetated deposits on the bluff. The oblique aerial photograph shows that in 1944, A largest lichen on a boulder near the exit of the the high level lake came to within about 30 m of the Rio Colonia at the southern end of Lago Colonia and summit of the terminal moraineŽ. Figs. 4 and 6 ; thus an oldest tree 65 m above the lake on top of the covering the fluvioglacial terrace in ca. 20 m of ancient terminal moraine date to 1896 and 1897, water. It also shows the Arco Glacier covering the respectivelyŽ. Table 1 . These dates suggest that ma- top of the recessional moraine and calving into the jor floods at the end of the 19th century may have lake. A tree on the moraine's distal flank demon- disturbed the boulders and overtopped the moraine. strates that the glacier lay behind it by 1956 and Glacier retreat during this century is traced by flood levels continued to decline reaching no higher several oldest trees and largest lichen at the northern than the middle of the beach by 1946. The last flood end of the lake. On a moraine fragment on the of any magnitude in the valley is dated by both trees eastern side of the outwash plain and 2 km from the and lichens below the terminal moraine to 1963 Cerro Negro outlet, there is a lichen with an axial Ž.Table 1 and Fig. 6 ; this agreement suggests that measurement of 420 mm representing an exposure any earlier tree or lichen growth was excluded by date of 1904 and to its northeast on a terrace 20 m long-standing water or frequent redistribution of de- above water level, a tree and a largest lichen suggest posits in the lower valley. exposure dates of 1914 and 1917, respectivelyŽ see The lichen population size±frequency distribu- Fig. 1 and Table 1. . On the opposite side of the tions measured on the boulders of the fluvioglacial valley and at the same height above water level terrace support the hypothesis of periodic catas- where a lateral moraine turns into the valley to trophic events. These populations typically display become an end moraine, another tree provides a colonization peaks followed by gaps and irregular bi- 1917 date for its top surface confirming glacier or multi-modal distributions, suggesting numerous retreat north of this position by the second decade of partial recolonization episodes with remote outliers the century. In 1944, the aerial photograph shows representing oldest survivorsŽ. Fig.5a . On the Arco there were two active outwash channels, but by Glacier foreland, there are three populations showing 1958, the western channel was dry according to similar but younger and less fragmented distributions TanakaŽ. 1980 . The date supplied by a tree bordering with mortality less evidentŽ. Fig. 5b . the dry channel bed suggests that it was probably already dry by 1953Ž. or possibly 1957, see Table 1 4.5. EÕents in the Arenales and Colonia Õalleys and that this position remained unaffected by any later outburst floods from the Arco. In 1944, how- At the junction of the Arenales and Colonia ever, the Colonia ice would still have been pressing Glaciers close to the ice-formed trimline, there is a against the Cerro bluff and a tree 27 m above water largest lichen of 472 mm growing on bedrock; this on its east side suggests this was the maximum ice measurement, divided by 4.7 plus 13 years delay height for that time; another at 3 m indicates a before colonization, provides a minimum date for the maximum height in 1964. trimline of 1883. The skewed curves of three lichen In the valley's centre on the outermost margin of population distributions on the highest of the four the recent moraine system, shown as the ice front on moraines below the trimline show a descending age both the 1975 map and the 1979 aerial photograph, a range, with the largest lichen in the youngest distri- 113-mm lichen, representing a date of 1969Ž or V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194 193

1975, see Table 1. , suggests there was a period of The possible errors in the estimated values for stillstand. Renewed retreat beginning in 1980 is indi- tree colonization are set by the 1958 and 1944 cated by three juvenile lichen size±frequency distri- floods. These provide a maximum possible coloniza- butions, with their largest lichens representing expo- tion delay of 22 years near water or 26 years for sure between 1980 and 1983Ž. Fig. 5d ; there is a trees on the fluvioglacial terrace beside Lago Arco. similar dateŽ. 1981 for the youngest lichen distribu- Although there is no direct support for the assumption on the most recent moraine bordering the tion of a 26-year colonization delay for trees on the proglacial lake on the Colonia Glacier's western Arco debris cone, the other evidence and the con- flank at the foot of the Arco valleyŽ. Fig. 5e . formity of dating with that from other sources suggest that the assumption is justified. Although there was no evidence to support a minimum estimate, 5. Discussion other worksŽ Villalba et al, 1990; Winchester and Harrison, 1994. suggest that a minimum colonization The major question addressed by this study is delay in optimum conditions could be 6 years, but dating confidence. Where dating is based on largest this minimum seems unlikely in the xeric environ- lichen sizes, statistical error terms cannot be pro- ments on the eastern side of the North Patagonian vided, but controls for the direction or size of possi- Icefield. ble errors is supplied here for both lichens and trees by flood dates, aerial photographs, and the regional 6. Conclusions dating described in earlier papersŽ e.g., Heusser, 1960; Warren and Sugden, 1993; Winchester and The work showed how dendrochronology and Harrison, 1996. . The main problem for the study was lichenometry can assist in the calculation of colo- the scarcity of dated reference surfaces for estima- nization and growth rates and together can provide tion of tree and lichen colonization and growth rates. dating controls for substrate exposure. The combined Necessarily, any between-area comparison of lichen use of these techniques supplied supportive dates for growth rates and colonization delays should be re- a sequence of events in the Arco and Colonia valleys garded with extreme caution due to micro-environ- over the last 120 years. The results from this study mental variations. Hence, the unexpectedness of the show dating agreement for surfaces likely to have identical 4.7 mmryear growth rates of P. patago- been exposed or affected by floods around nica and P. perrugosa, respectively, on the icefield's 1896r1897, 1914r1917, and 1963. In addition, aerial humid west and xeric east sides. The genus' known photographs support dating for the early 1980s posi- preference for damp conditionsŽ a preference rein- tion of the Colonia's terminus. Differences in tree or forced by the healthy appearance of P. perrugosa's lichen colonization delays or lichen growth rate growth around Lago Arco shores. indicates that it would exclude any such correlations. In the Arco and might be expected to grow more slowly in a drier Colonia valleys, the extent of the delay in tree environment. However, the close agreement of tree colonizationŽ. from 22 to over 90 years reveals the and lichen estimates across the dating range suggests potential scale of dating error if colonization delays that the species' growth remains approximately con- are incorrectly estimated. If tree age estimates below stant, at least at a distance from water. If growth is core height are also incorrectŽ varying here between faster on the shoreline, then the questionable dates 5 and 41 years. , then errors would be compounded. supplied here are those near water level. An increase A further measure of confidence in the selected in lichen growth rate near water from 4.7 to 6 colonization and growth rates for trees and lichens mmryear, a feasible increase based on rates found in was provided by the conformity of the minimum similar thallus forms elsewhereŽ V.W., unpublished 1881 and 1883 water and ice trimline dates, with the data. , would make a 3±6-year difference in the dates 1870s±1880s dating for the start of glacier retreat at of surfaces exposed in 1980 or 1969Ž. see Table 1 ; the end of the `Little Ice Age' on the west side of the this dating would still be compatible with the photo- icefieldŽ Winchester and Harrison, 1996; Harrison graphic evidence. and Winchester, 1998. . 194 V. Winchester, S. HarrisonrGeomorphology 34() 2000 181±194

The results from this study, apart from showing valleys, Hielo Patagonico Norte. Arctic, Antarct. Alp. Res. 32 the likely scale of glacial outbursts at the end of the Ž.1 , in press. `Little Ice Age' and subsequent declining levels of Heusser, C.J., 1960. Late-Pleistocene environments of the Laguna San Rafael area, Chile. Geogr. Rev. 50, 555±577. floodingŽ. Fig. 6 , focus attention on the importance Heusser, C.J., 1964. Some pollen profiles from the Laguna de San of detailed studies of colonization and growth rates Rafael area, Chile. Ancient Pacific Floras. University of Hawaii before dendrochronology and lichenometry can be Press, pp. 95±115. applied to the analysis and dating of geomorphic Matthews, J.A., 1992. The ecology of recently-deglaciated terrain. events. The wider significance of the study is that it A geoecological approach to glacier forelands and primary succession. Cambridge Studies in Ecology. Cambridge Univ. should produce more accurate dating of recent glacier Press, Cambridge, p. 386. fluctuations around the North Patagonia Icefield, Mercer, J.H., 1970. Variations of some Patagonian glaciers since where dated reference surfaces are extremely scarce. the Late-Glacial: II. Am. J. Sci. 269, 1±25. Nichols, R.L., Miller, M.M., 1951. Glacial geology of Ameghino Valley, Lago Argentino, Patagonia. Geogr. Rev. 41, 274±294. Ried, A., 1960. Stoffwechsel und Verbreitungsgrenzen von Acknowledgements Flechten: II. Wasser-und Assimilations-haushalt, Entquel- lungs-und Submersionresistenz von Krustenflechten benach- barter Standorte. FloraŽ. Jena 149, 345±385. We are grateful to the Linnean Society of London Schweingruber, F.H., 1988. Tree Rings: Basics and Applications and the University of Coventry for funding and of Dendrochronology. Kluwer Academic Publishing, Nether- CorporacionÂ Nacional ForestalŽ. CONAF for permis- lands, p. 276. sion to work in the San Rafael National Park. D.J. Shipton, E.A., 1964. Crossing the North Patagonian ice cap. Alp. Galloway kindly identified the lichens for us. Our J. 69Ž. 309 , 183±190. Sweda, T., 1987. Recent retreat of Soler Glacier, Patagonia as work in this unique environment was made possible seen from vegetation recovery. Bull. Glacier Res. 4, 119±124. by the organizational and logistical skills of Raleigh Tanaka, K., 1980. Geographic Contribution to a Periglacial Study International; we warmly thank all their hardworking of the Hielo PatagonicoÂ Norte with Special Reference to the personnel in London and at their field headquarters Glacial Outburst Originated from Glacier-Dammed Lago Arco, in Coyhaique. Exceptional support in the field was Chilean Patagonia. Center Company, Tokyo, p. 97. Villalba, R., Leiva, J.C., Rubulls, S., Suarez, J., Lenzano, L., provided by Dr. Ian Hughes from A.N.U. Canberra, 1990. Climate, tree-rings, and glacial fluctuations in the Rio Al Conner, Sandy Carmello, and Paul Barnes and Frias Valley Rio Negro, Argentina. Arct. Alp. Res. 22Ž. 3 , two teams of Raleigh Venturers. Finally, we thank 215±232. R. Naruse and M. Aniya for their help in tracing K. Warren, C.R., 1993. Rapid recent fluctuations of the calving San Tanaka, and M. Aniya for his helpful comments on Rafael Glacier, Chilean Patagonia: climatic or non-climatic? Geografiska Annaler 75AŽ. 3 , 111±125. an earlier draft of this paper. Warren, C.R., Sugden, D.E., 1993. The Patagonian Icefields: a glaciological review. Arct. Alp. Res. 25Ž. 4 , 316±331. Winchester, V., 1989. An evaluation of lichenometry: with field studies in Lappland, Britain and the Western Alps. D. Phil. References Thesis, Faculty of Anthropology and Geography, University of Oxford. Beschel, R.E. 1950. Flechten als Altersmasstab rezenter moranen.È Winchester, V., Harrison, S., 1994. A development of the licheno- Zeitschrift furÈ Gletscherkunde und Glazialgeologie, 152±161. metric method applied to the dating of glacially influenced Arct. Alp. Res., 5Ž.Ž 4 1973 . 303±309. debris flows in Southern Chile. Earth Surf. Proc. Landforms Beschel, R.E., 1961. Dating rock surfaces by lichen growth and its 19, 137±151. application to glaciology and physiographyŽ. lichenometry . 1st Winchester, V., Harrison, S., 1996. Recent oscillations of the San International Proceedings on Arctic Geology, 1960. In: Raasch, Quintin and San Rafael Glaciers, Patagonian Chile. Ge- G.O.Ž. Ed. , Geol. Arct. Vol. 2 Univ. Toronto Press, pp. ografiska Annaler 78AŽ. 1 , 35±49. 1044±1062. Wirth, V., 1987. Die Flechten Baden-WurtenbergsÈ Verbreitungsat- Harrison, S., Winchester, V., 1998. Historical fluctuations of the las. Verlag Eugen Ulmer, p. 528. Gualas and Reicher Glaciers, North Patagonian Icefield, Chile. Yoshida, K., 1981. Estudio geologicos del curso superior del Rio Holocene 8Ž. 4 , 481±485. Baker, Aisen, Chile. Science PhD, Geology thesis, Departa- Harrison, S., Winchester, V., 2000. 19th and 20th century glacier mento de Geologia y Geofisica, Universidad de Chile, Santi- fluctuations and climatic implications in the Arco and Colonia agoŽ. unpublished .