Biology of the Bamboo <Emphasis Type="Italic">Chusquea Culeou </Emphasis> (Poaceae: Bambusoideae) in Sout

Biology of the bamboo Chusquea culeou (Poaceae: Bambusoideae) in southern Argentina

Anita K. Pearson 1, Oliver P. Pearson i & Isabel A. Gomez 2 l Museum of Vertebrate Zoology, University of California, Berkeley, CA 94720, USA 2 Grupo de Andlisis de Sistemas Ecol6gicas, Bariloche, Argentina

Accepted 3.2.1993

Keywords: Biomass, Chusquea culeou, Nothofagus forests, Patagonia, Productivity, Bamboo

Abstract

Over a period of 7 years the biology and phenotypic variability of Chusquea culeou were studied at 5 locations in cool temperate forests of southern Argentina. Excavated rhizomes had an average of 1.1 successful rhizome buds, and an average of 2.1 years elapsed between successive generations of rhizomes. Rhizome buds usually develop within the first four years after a rhizome forms. Height, volume and weight of a culm can be calculated from its diameter 1 m above the ground. Culm size, length of foliage leaf blades, and pattern of secondary branching differed among study sites. Dead culms were numerous and commonly remained erect for more than 7 years after dying. New culm shoots appear in spring and reach full size within a few months. Shoots can grow more than 9 cm/day. Less than half of the shoots survived a year; most were killed by moth larvae. Multiple primary branch buds emerge through the culm leaf sheaths in the second spring. The mean number of branch buds at mid-culm nodes varied between 34.8 and 81.5, and the mean number of primary branches was between 22.8 and 40.8. Number and length of branches, and number and length of foliage leaf blades at each node is related to the position of the node on a culm. Most branches grow about 3 cm and produce 1 to 3 foliage leaves annually. Foliage leaf blades generally live 2 years or more; few survive 6 years. Relative lengths of foliage leaf blades and their spacing along a branch permit recognition of annual cohorts. Both gregarious and sporadic flowering have been reported, and every year a few isolated plants flower and die. Length of the life cycle is unknown. Seedlings require up to 15 years to produce culms of mature size. Foliage branches may live more than 23 years, and culms may survive 33 years. Extensive loss of new shoots to predation suggests that gregarious flowering may be driven by a need to escape parasitism. C. culeou clumps expand slowly. Average annual rate of increase of the number of live culms in a clump was 4.6?o. Methods of seed dispersal are undocumented. A dense stand of Chusquea culeou had an estimated phytomass of 179 tons/hectare (dry weight), 28 ?o of which was underground. Net annual production was about 16 t/ha dry weight.

Nomenclature: Voucher specimens from bamboo plants in this study were deposited in the following herbaria: Inst. de Bot~nica Darwinion, San Isidro, Argentina (SI); Iowa State University, Ames, Io- wa (ISC); Smithsonian Institution, Washington D.C. (US); and University of California, Berkeley (UC). Specimens, photographs and drawings of the monitored clumps were examined by Dr. Lynn Clark of Iowa State University, who identified all as Chusquea culeou E. Desvaux in Gay (1853), Hist. Chile, Bot. 6: 450. 94

Introduction 47°S in the southern cone of South America (Parodi 1945)• In Argentina, Chusquea culeou oc- Bamboos of the genus Chusquea are widely dis- cupies a narrow band along the eastern Andean tributed in Central and South America (Clark slopes of the Provinces of Neuquen, Rio Negro, 1989). However, unlike Asiatic bamboos that and northern Chubut; Parque Nacional Nahuel have been extensively utilized by native popula- Huapi lies near the center of its range, at latitude tions and have been cultivated and studied for 41°S (Fig. 1). Here in the park's glacier-carved hundreds of years (Janzen 1976), little use is made valleys are great native forests dominated by of Chusquea. It is not cultivated, and few studies beech trees of the genus Nothofagus, with bamboo have been published on any aspect of its biology as a major understory component• Although man (see Veblen 1982). has had relatively little impact on these forests, Our observations on Chusquea were begun in glaciers, avalanches, mudslides, volcanic ashfalls 1978 incidental to a study of small mammals in and fires have all influenced the distribution of the bamboo habitats in Parque Nacional Nahuel beech trees and the bamboo• Precipitation also Huapi, southern Argentina (Pearson & Pearson has had an effect: bamboo is more abundant in 1982; Pearson 1983). One goal was to document interactions between small mammal populations and the flowering cycle of the bamboo, but the lack of well documented information about the natural history and flowering behavior of the Chusquea, and uncertainties about its taxonomy encouraged us to focus on this plant. The abun- dance of Chusquea in many plant communities within the park (Mermoz & Martin 1987) offered an unusual opportunity to study the bamboo in its natural, relatively undisturbed environment. 1 From 1984 through 1991 we observed and mea- I \ sured selected Chusquea culeou plants in an effort I to document phenotypic variability and to quantify growth, productivity and survival. Similar detailed, long-term observations are not available for other American bamboos. We also wished to learn what we could about the flowering cycle. Like other woody bamboos, the bamboos of Parque Nacional Nahuel Huapi flower and die r., after a long vegetative period, but there are few

precise records of the life cycle of any species of ,a Chusquea. P arodi ( 1941, 1945) recognized three species of Chusquea in southern Argentina (C. culeou Des- vaux, C. argentina Parodi and C. montana Phil- ,a Verana a ippi), and in addition described the subspecies I I I I I C. culeou longiramea. All four were said to occur 0 20 km in a small western portion of Parque Nacional Nahuel Huapi (Parodi 1941, 1945; Nicora 1978), Fig. 1. Location of Parque Nacional Nahuel Huapi in south- but only C. culeou, the species we have studied, is ern Argentina. The five areas where clumps of C. culeou were widely distributed, occurring between 35°S and studied are indicated. 95

Table 1. Characteristics of study areas with marked clumps.

Locality Altitude Annual Exposure Light intensity index* Associates within 2 m precip. Periphery Periphery + 2 m

Puerto Blest (71 ° 49' W; 41 ° 1' S

Clump # 1 765 m 400 cm Dense shade 6.20; 6.3°o Nothofagus dombeyi Myrceugenia chrysocarpa Berberis linearifolia Clump #2 765 m 400 cm Dense shade 5.4°; 9.4°; N. dornbeyi

Castafio Overo (71 ° 47' W; 41 ° 12' S)

Clump # 1 870 m 200 cm Valley floor 32.40 95.1~; N. antarctica Some shade Clump #2 870 m 200 cm North slope 20.4°0 70.6°; N. dombeyi, N. pumilio, Blechnum sp., Berberis linearifolia, B. darwini, B. sp., herbs

Llao Llao (71 ° 33' W; 41 ° 3' S)

Clump # 1 780 m 240 * Dense shade 13.5 ° o 9.60o N. dombeyi Clump # 2 780 m 240 Dense shade 1.4°o 4.200 N. dombeyi

La Veranada (71 ° 28' W; 41 ° 29' S)

Clump # 1 1090 m ... Valley floor 34.0 o; 80.0 o5 N. antarctica, Berberis Some shade buxifolia, thick grass Clump # 2 1090 m ... Some shade 28.4?, o 68.4°0 N. antarctica, Ribes cucullatum, grass and herbs Clump # 3 1090 m ... Some shade 21.20o 74.4°0 N. antarctica, B. buxifolia, grass and herbs

Cerro Otto (71 ° 20' W; 41 ° 8' S)

Clump # 1 1145 m 130 * SE facing 29.5°0 100.0°0 Berberis sp., Ribes sp. slope; open Gaultheria sp., Ovidia pillo-pillo, grass and herbs Clump # 2 1160 m 130 SE facing 6.5) o 34.6°o N. pumilio slope; shade

* Estimated as described in Methods.

the wet, mild climate near the Chilean border Methods (over 4000 mm annual precipitation), and disap- pears toward the drier, colder eastern edge of the Over a period of 13 years we recorded our own park 60 km away (600 mm annual precipitation). observations and information gathered from local Chusquea grows and thrives near Lago Nahuel inhabitants concerning the vegetative character- Huapi (elevation 760 m), where most precipita- istics and flowering behavior of Chusquea culeou. tion falls as rain, but is also found up to 1450 m During six years, detailed measurements were along the ski slopes of Cerro Catedral. It occurs made on two clumps of bamboo at each of five both in pure stands in the open as well as beneath locations differing in climate, altitude and plant the dense canopy of Nothofagus forests. associations (Fig. 1 and Table 1). Although the 96

Table 2. Characteristics of monitored clumps.

Clump Year Clump Culms over 1 meter in height marked area a (m2 ) N N/m 2 Live:N Mean diam b Max N Mean height c Mass culms Mass culms Mass clump -+ se nodes + se - foliage d + foliage ° FW/m 2 (mm) (m) (kg) (kg) (kg)

Puerto Blest #1 1984 6.19 123 45.6 108(88~o) 20.25_+0.29 58 5.14_+0.07 ll2.1 138.2 22.3 Puerto Blest #2 1985 0.72 52 72.2 47(90%) 12.53+0.25 40 3.33_+0.06 15.7 20.4 28.4

Castafio Overo # 1 1984 6.42 121 18.8 107 (889o) 8.27 + 0.37 42 2.33_+ 0.09 17.2 25.6 4.0 Castafio Overo #2 1985 2.35 44 18.7 38 (869°) 5.73 + 0.40 36 1.74 + 0.09 4.0 8.2 3.5

LlaoLlao#1 1984 4.82 146 30.3 108(749o) 21.47_+0.69 44 5.43_+0.16 121.8 149.6 31.0 LlaoLlao #2 1984 7.36 269 36.6 203(75~o) 15.35_+0.26 39 3.99_+0.06 120.1 151.9 20.6

LaVeranada #1 1984 2.01 60 29.8 56(9390) 7.98_+0.34 33 2.27_+0.08 7.5 11.3 5.6 LaVeranada #2 1985 7.06 138 19.4 121(88%) 5.74_+0.20 32 1.74_+0.05 6.2 12.5 1.8 La Veranada #3 1988 10.52 189 17.2 180 (95~o) 8.12_+ 0.15 32 2.30 + 0.04 25.6 38.2 3.6

CerroOtto #1 1984 5.82 386 66.3 313(81~o) 18.49_+0.52 58 4.73_+0.12 259.4 322.0 55.3 CeroOtto #2 1984 5.48 234 42.7 195(83%) 17.07_+0.24 42 4.39+0.06 161.1 201.3 36.7

Calculated from measured circumference of clump 1 m above ground. 2 Based on measurements of all live culms except at Llao Llao # 1 (N = 32) and Cerro Otto # 1 (N = 54). c Calculated from diameters of live culms by regression: Height (m) = 0.234 x diam. at 1 m (ram) + 0.40. d Fresh weight (FW) of live culms without foliage calculated from regression: Mass (kg) = (10.68 x radius 2- 36.5)/1000. Radius (mm) measured at lm. e FW of live culms with foliage calculated from regression: Mass (kg) = (12.73 x radius: - 0.68)/1000. r Defined as the number of one-year-old culms over 1 m tall. g Counted on a minimum of 5 nodes in middle of foliage-bearing section of culms. h Calculations based on dry weight of litter samples collected under each clump, the original area of clump, and the number of live culms present in the clump in 1990.

pachymorph rhizomes of C. culeou result in a the ground. The averages of each of these two sets clump-forming growth habit, what appears to be are expressed as percentages of the light intensity a discrete group of bamboo culms may be derived of unobstructed sky measured at the same time from more than one seed. Throughout this report (Table 1). we use the word 'clump' instead of 'plant' to ac- All culms in clumps selected for study were knowledge this possibility. Bamboo clumps were marked with a permanent marking pen, and their selected for long-term study on the basis of the condition recorded (dead, alive, broken, < 1 m, following criteria: sufficiently separated from > 1 m). As young culms appeared in subsequent other clumps to suggest integrity as one plant, years, they were identified individually with a hidden or protected from disturbance by humans marking pen and with permanent plastic tags. and livestock, and with enough culms to suggest Measurements were made annually in spring, vigor and maturity but not an unmanageable when the previous year's production of shoots, number for the investigators. branches and leaves could be identified easily. Standardized measurements were made of Early in the study, diameters and heights of the each bamboo clump selected for study (Tables 1 culms were measured. Since height was found to and 2). Light intensity was measured in late No- be closely correlated with diameter of the culm at vember using a photographic exposure meter 1 m in all study areas, culm diameter became our aimed upward. Eight readings were taken around standard measurement, and culm heights were the periphery of the clump 30 cm above the then calculated from the regression between these ground, and another set of peripheral readings two measurements (Table 2). The diameter of was made 2 m out from the clump and 2 m above every culm was measured with dial calipers in the 97

Shoots Buds & Branches Litter h (DW)

Mean N Mean N successful Mean N Mean N &range Mean/clump/yr Mean/culm/yr Mean/m2/yr Mean litter shoots/yr shoots/yr r buds/node g branches/node g tons/ha/yr gm N leaves gm N leaves gm N leaves

13.8 4.8 49.2 40.8 (27-61) 696.6 12,720 5.8 105 258 4711 2.58 4.7 4.3 60.8 22.8 (13-29) 29.3 16.8 46.2 31.5 (24-40) 11.0 8.5 46.8 38.0 (25-47) 131.6 10,114 1.9 144 56 4304 0.56

20.0 8.1 81.5 38.4 (13-58) 477.2 33,679 3.8 178 99 4576 0.99 42.3 13.1 34.8 32.7 (18-54) 794.8 35,777 3.6 161 108 4861 1.08

24.2 11.8 10.8 8.2 38.2 27.3 (20-35) 628.3 37,503 4.7 282 89 5312 0.86 65.0 19.0 42.6 32.7 (24-36) 1009.9 65,592 5.6 364 96 6235 0.96

4.8 4.0 74.5 32.4 (16-56) 2677.2 102,822 10.8 413 460 17667 4.60 30.1) 13.6 49.4 37.0 (21-50) 948.0 31,932 4.6 154 173 5827 1.73

middle of the internode closest to 1 m above the the number of new, unmarked (yearling) new ground. A similar measurement was used in stud- culms. Rate of growth of shoots was calcu- ies of Chusquea culeou in Chile (Veblen et al. lated from repeated observations of tagged 1980). shoots. To determine culm volume, the diameter and Production, growth and survival of bamboo length of each internode was recorded on sample foliage was studied by repeated examination of culms from four study sites. Assuming each in- foliage leaves and branches identified with mark- ternode to be a cylinder, volumes of the culms ing pens or tags. Some foliage studies were con- were calculated by summing the volumes of the ducted on monitored clumps; others included internodes. nearby plants. Depending on the objective, mea- At each study site, culms from plants near the surements were made at intervals of one day to marked clumps were collected, measured and one year. stripped of their foliage. Culms and foliage were We placed three metal cans, 15 cm deep and weighed on a Pesola spring scale when fresh (FW) 13 cm in diameter, under each of eight monitored and again after being air dried at room tempera- clumps to determine the amount of litter dropped ture to a constant weight (DW). These data were annually by Chusquea culeou. The contents of the used to determine the relationship between culm cans were collected at yearly intervals and com- diameter and amount of foliage. ponents from Chusquea were separated and dried The previous year's production of culm shoots to a constant weight. Dried weights of the litter in a clump was determined each spring by samples and the number of foliage leaf blades adding the number of dead shoots present to included were converted to a measure of the an- 98

nual litter produced per square meter by each Study sites clump (Table 2). Samples of rhizomes were excavated from I. Puerto Blest (43 km WNW Bariloche) monitored clumps and from nearby bamboo The most western study site is in an old-growth plants at the conclusion of our study of above forest of 35-m-tall, evergreen, southern beech ground features in 1991. trees (Nothofagus dombeyi) that has been part of To obtain an estimate of the mass of culms the national park since its establishment in 1904 relative to the associated roots and rhizomes, we (Fig. 1). The main understory plant in the forest excavated three small sections of bamboo plants. is bamboo, which is protected from wind by the The culms were measured at 1 m, cut offat ground dense Nothofagus canopy but subject to bombard- level and weighed (with foliage) on a Pesola spring ment by falling branches and trees. There are also scale. The associated roots and rhizomes were some saplings and bushes (Table 1), but on the washed, separated and weighed while fresh and forest floor there is no grass and little herbaceous after drying at room temperature for one year. growth, due to the dense shade provided by the To quantify the amount of bamboo present, we high canopy of large N. dombeyi and the lower conducted a census through a dense thicket of canopy of leafy bamboo branches (described in C. culeou near the monitored plants on the Llao Ward 1965, and in Pearson & Pearson 1982). Llao Peninsula. A cord was stretched in a ran- Two bamboo clumps growing on well-drained, domly chosen direction for 36 m, and the diam- level land in this forest, and approximately 500 m eter at 1 m aboveground of every live or dead apart, were marked and monitored. culm within 1 m of the cord was recorded. The sample selected in this manner covered 71.6 m 2, 2. Rio Castaho Overo (44 km W Bariloche) or 0.00716 ha. This forest has never been lumbered or burned, Records of temperature and precipitation for but there is some livestock present. One marked Puerto Blest and for Pampa Linda (3 km SE of clump is above a 2 m road cut and further pro- the study site at Castafio Overo) were made avail- tected from grazing by fallen limbs from nearby able to us by the Grupo de Investigaci6n Ecol6g- Nothofagus antarctica, a deciduous beech. The ico, Parque Nacional Nahuel Huapi. Precipita- other monitored clump is a kilometer upstream tion for the Llao Llao study site was estimated on a north-facing slope dominated by large Notho- from the regression of mean annual precipitation fagus dombeyi. The area around this bamboo against distance (in km) from the Chilean border clump, opened up by the fall of several large trees, at seven localities (data from Boelke 1957). The contains 2 to 3 m shrubs, especially Berberis lin- regression is: earifolia. The general ecology of this region on the NE slope of Cerro Tronador is described by Gal- Precipitation (ram) lopin (1978), Garcia et al. (1978), and Pearson & = 3977-87.5X + 0.525X2; r = 0.99 Pearson (1982). Precipitation for the study sites on Cerro Otto, 380 and 395 m above the town of Bariloche, was 3. Peninsula Llao Llao (24 km WNW Bariloche) estimated by applying the altitude correction of Two clumps were selected in the level area north Gallopin (1978) to the mean annual precipitation of Lago Escondido. Although now primarily a at Bariloche (1065 mm, Boelke 1957). It should forest of evergreen Nothofagus dombeyi and Aus- be noted that annual precipitation from 1987 to trocedrus chilensis, this area was used for agricul- 1991 was below the ten-year mean. ture and grazing prior to 1936. What were once Data were analyzed by t-test, analysis of var- clearings in the forest are now pure stands of iance, or correlation analysis. Statistical compari- bamboo. The soil is a dark humus and in level sons were considered to be significantly different areas tends to remain moist. Plant associates at if p < 0.05. this site are limited by the denseness of the 99

Figs. 2-5. (2) La Veranada Clump # 2; (3) La Veranada Clump # 3; (4) Cerro Otto Clump # 1; (5) Cerro Otto Clump # 2.

Chusquea itself, which has succeeded in shading 4. La Veranada (43 km SSW Bariloche) out almost all undergrowth. One monitored clump Bamboo is the dominant understory plant in a is within 10 m of a large N. dombeyi; the other is widespread stand of Nothofagus antarctica at this in a previously open area that is now all Chusquea. location. The beech trees are uniformly about 40 Here, as at Puerto Blest, the Chusquea is pro- years old and 7 m high. They have grown from tected from wind by the surrounding forest, but is root sprouts, which suggests widespread destruc- subjected to falling branches. tion of an earlier forest, probably by fire. N. ant- 100

arctica is deciduous, and forms an open canopy each node of the rhizome proper holds a single over the bamboo, shrubs, and herbaceous plants. large bud with the potential to develop into an- There are small grassy clearings among the thick- other rhizome. The length of mature rhizomes of ets of trees and bamboo. One bamboo clump C. culeou varied between 5.5 and 15.5 cm, includ- chosen in 1984 was destroyed in a roadbuilding ing the neck. Small culms and culms on young operation in 1987. Another clump was then cho- plants have shorter rhizomes with fewer neck in- sen. The two surviving clumps (Figs. 2 and 3) are ternodes and with fewer buds. 50 m apart and each is separated by several The lateral buds on a rhizome usually develop meters from other bamboo clumps. Saplings of basipetally, and often are in different stages of N. antarctica grow in the light sandy loam within development. The diameter of a young rhizome the periphery of the monitored bamboo clumps. increases after the apical culm has been produced. The vegetation at La Veranada is quantified in This increase is not dependent upon the apical Pearson & Pearson (1982). culm, however, for rhizomes terminating in short dead shoots also become enlarged and produce 5. Cerro Otto (2.7 km WSW Bariloche) further generations of rhizomes. Resources re- Two monitored bamboo clumps are located in sponsible for rhizome enlargement thus must rich loamy soil on the SE slope of Cerro Otto, a come from older parts of the plant. Rhizomes at 1500 m mountain adjacent to the town of Bar- the base of yearling culms have root primordia iloche. One clump is in a previously forested area but few developed roots; root development gen- that was partially cleared for pasture about 40 erally occurs in the second and subsequent years. years ago (Fig. 4). The clump has low shrubs, In forested areas, the rhizomes and roots of annuals and grass around its base, and is sepa- C. culeou form a complex mat within the top rated from other large bamboo clumps by more 20 cm of soil, below which are the roots of nearby than 3 m. About 100 meters away, the second trees. In treeless areas, the rhizomes form a three- marked clump of bamboo is in a forest of second- dimensional tangle that may be as deep as 30 cm. growth Nothofaguspumilio, a deciduous species of As rhizomes emerge from rhizomes, the relation- southern beech (Fig. 5). One tree (30 cm DBH) is ships between successive generations ofrhizomes located within the periphery of the bamboo clump. are easily followed. A new generation of rhizomes is produced each year in some species of bam- Results boos (e.g., McClure 1966, Fig. 2). In our monitored bamboo clumps we excavated 41 pairs of Rhizomes successive generations of rhizomes. From the recorded years of emergence of the apical culms, Rhizomes of C. culeou are short-necked and pa- intergenerational periods were found to vary from chymorph, with short, thick and solid internodes one to four years, with a mean of 2.1 years. Anal- (according to the classification of McClure 1966). ysis of these rhizomes suggests that bud develop- We did not find any leptomorph rhizomes. Each ment on a rhizome is restricted to the first four underground, horizontal rhizome ends apically in years following formation of the rhizome; none of a culm shoot, or if the shoot is successful, in an 25 rhizomes four years and older had produced upright culm. Culm shoots emerge above ground new shoots in the year they were excavated. Av- in the spring and are evidence of the completed erage production of these older rhizomes that had formation of an equal number of rhizomes under- presumably completed production was 1.1 rhi- ground. Each rhizome is connected to its prede- zomes per parent rhizome. Although the data are cessor rhizome by a slender rhizome neck of 8 to limited, they suggest a low rate of success in pro- 14 internodes lacking roots and buds (Fig. 6). The duction of new generations of rhizomes. Edaphic two to six internodes (most frequently four) of the factors and climatic factors with annual varia- rhizome proper give rise to roots and rootlets, and tions must affect these numbers. 101

'89-'90 '90-'91 '90-'91 '87-'88 '87-'88

I I 25 MM ffl '84 OR '85 '~000 '90-'91 ~C)

/ / '84-'85.

~-'89

///f Fig. 6. Diagram ofrhizomes excavated from La Veranada Clump # 3 (left) and Llao Llao Clump # 1 (right). The spring in which the culms emerged is indicated.

Rhizomes excavated late in the spring often ground biomass in the Asiatic bamboo Bambusa had elongate buds, although the majority of new arundinaria (Gadgil & Prasad 1984). A linear re- shoots are aboveground at this time. This obser- lationship exists between the mean diameter of vation, in addition to the intergenerational gaps of the culms and the underground biomass per culm several years between culms of known age, sug- in the three samples: larger culms arise from larger gests that rhizomes form slowly. A rhizome may rhizomes. If the underground biomass is divided give rise to two new rhizomes and their apical by the mean diameters of the culms in the three shoots in the same year, but it may require sev- groups, there is an average of 15 gm (FW), or eral years of underground development to do this. 9 gm (DW) of underground rhizomes and roots Excavated sections of three bamboo plants for each millimeter of culm diameter. weighed a total of 20 kg (FW) and included 32 Rhizomes outweigh roots about 6:1 when fresh, culms. Rhizomes and roots weighed an average 3.4:1 when dry. Approximately 40 ~/o of rhizomes 41~o (FW) or 39% (DW) of the aboveground and roots is water. mass, or about a third of the total biomass. Al- In summary, although there may be as many as though based on a limited sample, these figures six buds on a rhizome, some buds fail to develop, are in agreement with the proportion of under- many others begin development and die, leaving 102

an average of slightly more than one successful ened, sometimes green, basal internodes, and are rhizome bud for each mature rhizome. Insects more durable. Some stunted culms marked by us and small mammals contribute to the mortality of died within a year; others remained erect, leafless, the buds. Rhizome buds develop within the first and green at the base for more than six years. four years of the parent rhizome, and some re- There were between 44 and 386 mature culms quire several years of underground development over 1 m in height in the clumps we monitored before their apical culm shoots .emerge above- (Table 2). The culms were clustered in areas of ground. 0.72 to 10.52 m 2 (diameters of approximately 1-4 m). These figures are within the range of mea- Clumps surements reported for Chusquea culeou at a similar latitude and elevation in Chile (Haverbeck In spring (November and early December), the 1983), although Veblen etal. (1980) reported aboveground composition of a mature Chusquea smaller clumps of 30-50 culms covering an av- culeou clump includes: (a) live culms with foliage, erage area to 2 m in diameter. The mean above- (b)dead culms, with or without remnants of ground biomass of ten clumps in a Chilean forest branches, (c) live one-year-old culms without fo- (162 kg DW; Haverbeck 1983) is almost three liage (yearlings), (d) stunted culms (live or dead), times the mean of the clumps in this report lacking foliage and with parasitized or broken (56.9kg DW), reflecting primarily the greater tops, (e)short dead shoots from the previous culm size in Chile. spring, and (f)short, recently-emerged shoots (Fig. 7). Short dead shoots (10 to 15 cm) do not Culms remain intact through a second year, but stunted culms (from parasitized or damaged shoots that New culm shoots usually appear in spring or early failed to achieve a mature height) possess hard- summer, and growth of the culm (diameter and

24- 3 1 Culms )lm: 4 4 .(lm: 4 5 20 ¸ Dead 4 6 [--I Live E -516, 0 @ @ 7 0 7 L12. 5- -- 5 E z~ 8. .

O F-T-q 6 8 10 12 14 16 18 20 22 24 26 28 30 ABC Diameter of culm at lm (ram) Fig. 7. Composition of Puerto Blest Clump # 1 in spring, Dec. 4, 1990. All standing culms were counted; some culms dead in 1984 were still present. Numbers within columns indicate age in years of culms marked by us. A: Short, dead shoots (10-15 cm) from previous spring; B: Culms < 1 m in height (most are stunted by parasites and all lack foliage); C: Short, live shoots less than one month old. 103

height) is completed within the next three months. have more internodes. The mean height of mature A shoot that emerges late in the summer is ca- culms in our study clumps varied from 1.7 m at pable of maintaining its capacity for growth Castafio Overo and La Veranada to over 5 m at through the winter and into a second growing Puerto Blest and Llao Llao (Table 2). A few culms season, but this rarely happens. At the beginning achieve a height of 7 m. The maximal number of of their second year, young culms (yearling culms) nodes varied correspondingly (32 to 58; Table 2). lack foliage and are easily identified by the unin- Mid-culm internodal segments are longer than terrupted series of pink, green or purple culm those near the base or tip (Fig. 8A), as in Asiatic leaves that emerge at the nodes and surround the bamboos similarly analyzed (McClure 1966). Av- succeeding internodes. As the second spring eraged measurements of the sequence of intern- progresses, branch buds push through the base of odal lengths in C. culeou produce similar convex the culm leaves, and the first crop of foliage leaves curves for culms from four study sites and a appears. The slender blades at the tips of the culm somewhat different curve for culms from La Ve- leaves fall within their first year, but the lower ranada (Fig. 8B). The curve produced may be section of the culm leaves (culm leaf sheaths) may characteristic of a species (Porterfield 1933). remain attached for five or six years. Maximal diameter of a culm is determined by The height of a culm is determined by the num- the diameter of the rhizome from which it devel- ber and lengths of the internodes; taller culms ops, and internodal diameters decrease uniformly

150 170 o o oo o o o ° 150 t o A 130" /~~~p a o • a "~ uerto Blest 130 x~ "°° °~i ° o 110- O~ 110 • • i o ~:D 90- ~' 9o ~ 70- 70 50 Le VeranQdo .__m

30 tto 30-

A 10 I I I I ~ I I i ~ 5 I0 15 20 25 50 35 40 5 I0 15 20 Z5 50 ,55 40 45 50 ]nternode Number (from base) Internode Number (from bose)

30 , O 50

:5- ~ro Otto ~%13o 0) 20 °~oo o

o 15 AA~ ~I *o cD /5 ~ 10 •AAA °oo o

5 Cer "•" ""

0 I I I I I I ~'1 ~"'~1 ) 5 10 15 20 25 30 35 40 45 5 I0 15 20 25 30 35 40 45 50 Internode Number (from base) Internode Number (from bose)

Fig. 8. Internode length and diameter plotted against internode number. A and C show data for five culms from Cerro Otto with their average regression; B and D present regression lines derived from similar data for all study sites. (A) Cerro Otto: r = + 0.88; (B) La Veranada: r = + 0.82. Llao Llao: r = + 0.74. Puerto Blest: r = + 0.79. Castafio Overo: r = + 0.87; (C) Cerro Otto: r = - 0.96; (D) La Veranada: r = - 0.89. Llao Llao: r = - 0.97. Puerto Blest: r = - 0.99. Castafio Overo: r = - 0.98. 104

with increasing distance from the culm base 1400. Puerto Blest * •,, Llao Llao • (Figs. 8C and 8D). Culms of C. culeou in Chile 1200. La Veranada o have larger diameters and are taller (Haverbeck ~EoEIo00- Cerro Otto • 1983); the mean height for culms growing in a _E 800. Chilean forest (6.81 m) approaches the maximal height observed by us in Parque Nacional Nahuel Huapi. ~ 400- "5 In addition to its close correlation with the > 200- height of a culm (Fig. 9), the diameter of a culm 0 measured at a height of 1 m is tightly correlated 0 20 40 60 80 1 O0 120 140 with both its volume and its weight, with or with- (Radius at 1 meter) 2 out foliage. By using the square of the radius at Fig. 10. Relationshipof volume of a culm against the square of its radius (in ram) at 1 m above the ground, r = + 0.99 1 m instead of the diameter, these regressions be- (Y = 8.91X - 41). come linear (Figs. 10 and 11). For 10 fresh culms that were weighed and volumes calculated, the mean density was 1.25 + 0.02 grn/cm 3 (consider- and Puerto Blest, in part because there are more ably heavier than water). Dry weight of culms dead culms standing in these clumps. with foliage is 58~o of the fresh weight. Branches and leaves are 12~o to 50~o of the The total fresh weights of all live culms and fresh weight of the culm. Culms of smaller diam- their foliage in the monitored clumps (calculated eter tend to have a greater percentage weight of from the regression provided in Fig. 11) fall be- leaves, while culms with larger diameters have a tween 8.2 kg for Castafio Overo Clump # 2 and lighter percentage leaf load (Fig. 12). 322.0 kg for Cerro Otto Clump # 1 (Table 2). In Mature culms do not increase in diameter. In general, culm measurements from bamboo Cerro Otto Clump # 2, where over a hundred clumps at Llao Llao, Cerro Otto and Puerto Blest mature culms measured in 1984 were measured are similar, and contrast with those from Castafio again in 1990, the mean diameters were not sig- Overo and La Veranada. In the former group, nificantly different between the two years (t-test). culms are larger in diameter and taller, and sub- There were also no significant changes in the mean sequent calculations of the total mass of live culms diameters of yearling culms produced at each of are correspondingly large. Culm densities (culms/ the eleven monitored clumps during the six years m 2) likewise are greater at Llao Llao, Cerro Otto we recorded these measurements.

Puerto Blest * 1.2- PuerLo Blest * 6- Castano Overo A • Castano Overo& • J Llao Llao• * . v Llao Llao• • • J La Veranada o • • La Veranada o /" ~, 5 Cerro Otto • ~- Cerro Otto • _ • + 0.8-

u E0.6- 3 ~o.4- o~A*. -~ 0.2- • •

1 I I I I I I I I LL. 0.0 I I I I I I I I I 7 9 11 13 15 17 19 21 23 0 10 20 30 40 50 60 70 80 90 Diameter of culm at 1 m (in mm) (Radius at 1 m) 2 Fig. 9. Relationship of height of a culm against diameter Fig. 11. Relationship of fresh weight of a culm with foliage measured mid-internode closest to 1 m above the ground. against the square of its radius (in mm) at 1 m above the r = + 0.91 (Y = 0.23X + 0.40). ground, r = ÷ 0.97 (Y = 0.007 + 0.013X). 105

60- bearing section of the culm (in this case, nodes Puerto Blest * o Castano Overo & 18-22) is the most productive area and provides ~ 50 o & Llao Llao • A o La Veranada o the optimal material for comparison with other ~ 40 o Cerro Otto • plants. E ~, 3O

~ 20 Culm survival

-=:10- Young shoots have a high mortality, but when a culm shoot has completed growth at the end of its I I I I I I I 6 8 10 12 14 16 18 20 first summer and has hardened, its chance of sur- Diameter of culm at 1 m (in ram) vival for the next 6 years is high (Fig. 14). Of 28 Fig. 12. Relationship between diameter ofa culm at 1 m above culms marked as yearlings in 1984, 23 were alive ground and the percent of its fresh weight that is made up of in 1990 at seven years of age. Assuming that the leaves and branches, r = - 0.62 (Y = 47.6 - 1.62X). rate of survival continues as a straight line, ex- trapolation suggests that culms live for 33 years. Description of a culm Dead culms are a significant component of A single culm typical of the mature culms of bamboo plants. An average 16.2~o of the culms nearby plants was selected for detailed analysis. in the monitored clumps were dead at the time of The culm, estimated by a count of leaf cohorts to the original census (Table 2), and during our study be six or seven years old (technique discussed the percentage of dead culms increased to over a later), was collected at Cerro Otto on Novem- third in some clumps (Table 3). The situation is ber 13, 1988, before the new spring leaves ap- similar in Chile, where up to 25~o of the culms in peared. It had 31 nodes, a height of 3.46 m, and a clump were dead (Haverbeck 1983). a diameter at 1 m of 14.7 mm. Diameters at mid- Dead culms are remarkably enduring: although internodes varied from 19.2 mm at the base of the their foliage leaf blades were lost within a year, culm to 1.5 mm in the terminal internode. The culms recorded as dead in the Puerto Blest for- volume of the culm was 413 cm 3. When dried to est in 1984 remained standing and with some a constant weight the stripped culm weighed branches for more than seven years. A number 282 gm and the foliage 108.5 gm, producing a total of dried culms of a clump that flowered and died dry weight of 390.5 gin. at Puerto Blest in 1977 remained erect 13 years Live branches of the culm carried a total of later. 1468 foliage leaf blades, an average 3.41 blades per branch, with a range from 0 to 6 (a few young, Shoots live branches had no leaf blades). The average Culm shoots begin to sprout from the tips of un- length of the foliage leaf blades was 71.49 mm derground rhizomes in early spring (late October (range, 8 to 110 mm). Regression of dried weight and early November), and although an occasional of foliage leaf blade against blade length was de- shoot appears as late as February or March, 75~o termined from 82 of these leaf blades to be: Dry of a season's production has emerged by mid- weight (mg) = 3.4- 0.019X + 0.0047X 2, where December. X=length of the foliage leaf blade in mm; Shoot production varied considerably among r = + 0.94. From this, the total standing crop of the monitored clumps (Table 2). Cerro Otto the 1468 foliage leaf blades on the culm can be Clump # 1 produced between zero and three estimated to weigh 38.2 gm (DW). shoots in four of seven years. At the other ex- A detailed graphic profile of foliage character- treme, La Veranada Clump #3 produced 82 istics of this culm (Fig. 13) demonstrates the vari- shoots in one season. Clumps that produced nu- ability of all parameters at different nodes and merous shoots, did so every year, and those that suggests that the middle third of the foliage- produced few shoots likewise were consistent in 106

I 50. A 50 B 50 C

25 i 25 251, jl I f I I I I _4 I I I © 20 20 i 2O f I I 8 I I 15 15 15 I @ £ 10 10 105 t ~ Dead branches Total branches

O/ I I I I I-- 0 I I I I 0 - I I I I 0 10 2O 50 4O 50 0.0 2.0 4.0 6.0 8.0 0 5 10 15 20 Number of branches Total length live branches (m) Ave. length live branches (cm)

] 50 D 5o E 50 F I

25 I I 25 25 I I I I I e20 2o I 20 I ~D I I E J I I I c15 15 15 I © X3 21o 10 10

i I I 0 - I I i I 0 - I I I I 0 50 100 150 0.0 1.0 2.0 3.0 4.0 40 50 60 70 80 Total number of leaves Total dry weight leaves (gin) Average length leaves (ram)

Fig. 13. Vertical distribution of the standing crop of branches and leaves (foliage leaf blades) at each node of a six-year-old culm from Cerro Otto, Nov. 13, 1988. Primary and secondary branches are treated equally.

their low production. Shoot production depends between generations ofrhizomes is approximately on the number of rhizomes under four years of two years. age, regardless of whether they terminate in a Short new shoots elongate at a rate between dead shoot or in a live culm. A nonparametric 0.5 to 1.5 cm/day; taller shoots have a higher rate runs test of shoot production based on 34 clump- of growth (Figs. 15 and 16). Since average growth years of data showed that a clump tends to pro- rates of individual shoots over intervals of 21 to duce a larger crop of shoots in alternate years 58 days are between 5.0 and 9.6 cm/day, the ac- (p<0.05). This pattern supports the conclusion tual rates must be considerably higher at times. A derived from marked culms that the average time mean growth rate of 10.1 cm/day was the high- 107

loo•

5 • ...... o, 8ot~ :i t\ 4 /. • ...... - '~3v~'ff :.:;.!!i: .:.., : "I{.:: .o ......

0~-~--0-- O~ 0 . • --0~ O ..... 40 ,..,. 4~!:. >.. g. 2

0 =ii:~ ' ', I i t I I I I I I I Nov Dec Jan Feb M(~F Apr -:Io 1 2 3 4 5 6 7 Years of age FiE. /3. Growth of selected shoots at Llao Llao, 1984-85. Fig. 14. Survival of 1435 shoots and resultant culms over 1 m in height. shoot survival, each expressed as a percentage of est recorded for C. culeou in the more mild Chil- the number of live culms in the study clump in the ean climate (Haverbeck 1983). same year. Calculations were based on deviations of the monthly means in each year from the 10- Shoot survival year monthly means of precipitation and tem- Among our monitored clumps, the mean number perature recorded at the Pampa Linda weather of successful shoots (i.e., shoots that became station near Castafio Overo. It was assumed that yearling culms over 1 m tall) produced annually deviations at other study sites would be similar. was 7.2 ~o of the number of live culms in a clump, or, stated differently, there was a ratio of 14 live Production versus temperature. In the seven years culms to every yearling culm in a clump. Fewer in which shoot production data were collected, than half the shoots produced were successful the production of new shoots was significantly (46~o of 1435 shoots); many died before they correlated with the mean temperatures of two were 15 cm high. Over 40~o of the 673 dead months, December and June. Above average De- shoots examined were killed by insects. This is a cember temperatures were associated with a conservative number, since decomposition pre- greater number of shoots in that season cluded diagnosis of many shoots. (r= + 0.86); below average June temperatures Shoot survival in C. culeou varied among were associated with greater shoot production in clumps and among years, but in all clumps there the following spring (r = - 0.68). was poor shoot survival in the spring of 1983-84, when only 11 ~o of 236 shoots survived to become Survival versus temperature. Cool temperatures yearlings over 1 m in height. More shoots sur- from October through January were followed by vived in clumps that produced many shoots than better than average survival of shoots. The re- in clumps that produced few shoots, but percent- gression of mean temperatures of November and age survival of shoots was low in plants that pro- December considered together against the per- duced many shoots and higher in plants that pro- cent survival of shoots had a value of r = 0.79, duced fewer shoots (r = - 0.76 for 11 plants, 1435 p< 0.05; for the entire growing season (October shoots). Such a relationship would prevail if in- through March), r = -0.75, p = 0.05. Similar re- sect parasitism on the shoots were density depen- gressions for the winter months May through dent. September were all positive, but not significantly We examined the relationship between two cli- so. Thus, a colder winter and a warmer spring matic parameters, precipitation and temperature, produced more shoots, but a greater percentage and two measures of the productivity of the study of the shoots survived following warm winters clumps: annual shoot production and annual and cool springs. 108

Table 3. Repeated censuses of monitored clumps.

Clump Culms over 1 meter tall Mean )o change a Total Live culms N N ~o Live ~o Young ~o Yearling

Puerto Blest # 1 1984 123 108 88 6 1987 149 126 85 24 < 4 yrs 6 1990 148 121 82 28 < 7 yrs 2 + 1.9 #2 1985 52 47 90 13 1987 62 46 74 33 < 3 yrs 2 1991 74 50 68 50 < 6 yrs 10 + 2.9

Castafio Overa # 1 1984 121 107 88 3 1989 215 186 86 51<6 yrs 8 + 10.3 #2 1985 44 38 86 21 1990 85 70 82 70<6 yrs 16 + 11.9

Llao Llao # 1 1984 146 108 74 0 1989 166 124 75 43 < 6 yrs 8 + 2.8 #2 1984 269 203 75 2 1990 336 222 66 38<7 yrs 4 + 1.3

La Veranada # 1 1984 60 56 93 4 1987 98 92 94 43 <4 yrs 9 + 16.2 #2 1985 138 121 88 9 1989 164 133 81 22 < 5 yrs 7 + 2.4

CerroOtto # 1 1984 386 313 81 2 1989 396 249 63 8<6yrs 0.4 -4.6 #2 1984 234 195 83 3 1990 321 206 64 38<7 yrs 3 + 1.3 a Mean annual percent change in the number of live culms, based on prorated number of live culms for each year.

Production versus precipitation. Overall shoot ters, wet cool springs and a dry summer. It is production was significantly better in a dry likely that optimal conditions for shoot survival November-December (r = - 0.89, p < 0.01) and a were those that suppress the shoot parasites. wet January-February (r -- + 0.84, p < 0.02). Forage Survival versus precipitation. The precipitation regime that correlated with good shoot survival Buds and branches was the reverse of that for good shoot production: When a culm is 9 to 13 months old and a second shoot survival was better in a wet November growing season is approaching, primary branch (r= + 0.74, p= 0.05) and in a drier January buds at the nodes begin to enlarge beneath the (r = - 0.82, p = 0.02). culm sheaths. All plants in our study produced In summary, shoot production was increased by multiple buds at the nodes (Fig. 17), a character- cold winters, dry warm springs and a wet sum- istic of the genus Chusquea (Clark 1985, 1989). At mer; shoot survival was enhanced by warm win- some nodes, buds were small and numerous; 109

150 branches occurred on the middle nodes of the "•" foliage-bearing section of the culm (Figs. 13A, 120 o 13B, 18C, 18D). For five years we followed branch bud devel- ~ 90 oO "6 opment and branch growth on culms of known 60 o o age at Castafio Overo. Branch buds developed at -6 successively lower nodes during the first six years .~ 30 of foliage production, but not thereafter. In the first two seasons of foliage production, there was I I I I I I 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 a maximum of 25 branches per node; by seven Growth in 24 hours (cm) years of age some nodes had as many as 54 Fig. 16. Growth rate of shoots of different sizes plotted branches. By four years of age (i.e. after three against their initial heights. Measured at Castafio Overo, De- years of foliage production), some branches had cember 10-11, 1990. The mean temperature for the 24 h of observation was 7.5 °C. r = + 0.94 (Y = 1.78 + 9.4X + 9.4X2). died, and the first secondary branches appeared, developing either at the base or more distally on primary branches over 10 cm long. By eight years, other nodes had one to three slightly larger buds one-third of the branches were dead, another third among numerous small subsidiary buds. Basal had secondary branches, and three or four larger nodes had few buds, sometimes only one. We branches were present among the smaller counted between 32 and 96 buds at mid-culm branches at a few nodes. These larger branches nodes; in monitored bamboo clumps, the mean (presumably arising from the larger buds at a number of buds varied between 34.8 and 81.5 node) grew primarily in their first year. Elonga- (Table 2). tion of these and other primary branches was Although branch buds develop first at nodes at curtailed with production of secondary branches. the top of the culm, branches usually form at all Some branches grew 12 to 18 cm in a single nodes in the top third of the culm during the first growing season, and others only 1.5 cm. Most year of foliage production. The position of the (including new branches) grew about 3 cm a year lowest branch-producing node in young culms while putting out a new crop of leaves. At Castafio varies with the location of the plant. At Llao Llao, Overo the maximum recorded growth of a branch two-year-old culms had branches at all nodes in a single growing season was 22 cm; another above the 15th to 20th nodes; at Puerto Blest, branch grew 18 cm one year and 12 cm the next branches appeared only above the 24th node on before putting out a secondary branch. culms of a similar age. The greatest production of There were clear differences in the production

Fig. 17. Examples of mid-culm bud complements from Cerro Otto (left) and La Veranada (right). 110

of secondary branches among monitored bamboo Table 4. Annual leaf production and standing crop per live clumps, even within the same study site. In La branch. Veranada Clump #2, secondary branches Number of Range N a N b emerged from the base of 46 of 90 primary leaves branches examined; in Clump #3 there were (mean _+ se) fewer secondary branches (12 on 85 branches), and most arose from the second node on the Leaf production From cohort count: primary branch. The two branching patterns re- Puerto Blest 2.64 + 0.06 c 1-4 29 121 sulted in a different appearance of the two plants. Castafio Overo 2.37 +_ 0.04 1-4 61 208 In the former, the foliage appeared erect and at a Llao Llao 2.34 +_ 0.04 1-4 71 300 narrow angle with the culm (Fig. 2). In the latter, La Veranada 2.32 + 0.03 1-3 107 438 the foliage appeared more lax and formed a wider Cerro Otto 2.33 + 0.03 1-3 113 338 angle with the culm (Fig. 3). From marked leaves: Of 164 live branches examined from culms of Llao Llao 1.19 + 0.05 0-3 365 Cerro Otto 1.82 + 0.03 c 0-5 638 Cerro Otto Clump # 1, 45~o had secondary branches. A similar sample from Cerro Otto Standing crop in spring Clump # 2 displayed more complex branching Llao Llao 4.18+0.07 ~ 0-13 799 patterns, and 78~o of the branches had second- La Veranada 4.67 + 0.14 c 1-11 186 ary branches. A few branches from each of these Cerro Otto 3.93 + 0.08 c 1-6 204 clumps had fifth-order branches. Number of branches analyzed. The longest branch recorded, on a plant at Pu- b Number of cohorts counted. erto Blest, was 106 cm in length and had a total ° Differs significantly from others in this category. of 55 leaf blade scars and six foliage leaf blades. When analyzed by cohorts (justified in a subsequent section), branches from Puerto Blest Clump # 1 gave an estimated average annual foliage leaf Foliage characteristics of a culm vary with the production of 2.64 leaves/branch (Table 4). Thus age of the culm as well as with node number a branch with 61 blade scars and leaf blades (Fig. 18). A four-year-old culm had the greatest would be 23 years old. Six selected long branches number of live branches and foliage leaf blades from Cerro Otto Clump # 1 averaged 76 cm in per node (Figs. 18A and 18C). The reduced num- length and had an average of 39 leaf blade scars ber of live branches on an older culm (Fig. 18C) and leaf blades, which, divided by the mean an- reflected increasing mortality of branches with nual production of leaves per branch on Cerro age; the average length of live branches and total Otto, derived from marked leaves (1.82 leaves; leaf production per branch, however, was greater Table 4), indicates a similar branch longevity of on an old culm (Figs. 18E and 18F). The number 22 years. of foliage leaves at a node increased through an On a carefully analyzed culm from Cerro Otto, increase in the number of branches rather than by the most branches and the greatest total length of more live foliage leaves per branch. branches were produced by nodes 15 to 25 of a There were generally fewer branches per node total of 31 nodes (Figs. 13A, 13B). Of a total of on mature culms at La Veranada and Castaflo 524 branches on this culm, 18~o were dead, and Overo than at the other three study sites (Table 2). 15 Yo of the live branches had one to four second- If the number of primary branches per node on ary branches. The length of the live branches (pri- mature culms is compared with the number of mary plus secondary) ranged from 2.0 to 41.0 cm, branch buds present on young culms, it seems with an average length of 17.43 cm and a total clear that differences in branch numbers result cumulative branch length of 75 m for the whole from loss of buds or their failure to develop rather culm. than from a shortage of buds. 111

6.0 280 yye~; o,ldd cu,m :.: J~ ~/ A g B 2~ 5.0- $ 240 over 7 yeers old • • ~/ \/\ o E c~ 200 ~ 4-.0- o /o, > 160 > Z 3.0- o °X / ". ' " " "b 12() i 2.0- ; 80 c E X - • "\ ~ 40 ® 1.0- z ..o " o F 0 0.0 I I I I I I I 20 24 28 32 36 40 44- 16 20 24 28 32 36 40 4-4 Node number Node number

8 7 D 50- "\/\f\\ C ¢) E 6 o 40- v E 2 ~5 z~ / \ >~ 50- o 4 'S . ! 20- ...... \ X2 ~ 2 F • o\ u 10 z

• °X. Fa_~ / *. . " ~\ 0 T " ' " 16 20 24 28 .32 36 40 44 20 24 28 32 36 40 44 Node number Node number

25 • . ~1 15 ' F • g • , e • 20 12 J:_

o 19 z~

/ "* =10 /" > 6- E \[A ° , >.y-<...... o .... • • o i • ...... •• • • A ~ 3 - ~

l A < 0 I I I I I I I ~- 0 i I I I I I i 6 20 24 28 32 36 40 4-4 16 20 24 28 32 36 40 44 Node number Node number

Fig. 18. Analysis of six characteristics of the live foliage on three culms of different ages collected November 1 l, 1986, at Llao Llao. The term leaf here refers to foliage leaf blades. Branches includes both primary and secondary branches.

Foliage leaves the ligule of the terminal leaf. The furled leaf blade Foliage leaf blades develop furled within the pro- is propelled by growth of the leaf sheath to which tective covering of the terminal leaf sheath on a it is attached and by elongation of the branch branch. The new leaf blade achieves nearly its full itself as it forms a new node. Three to twelve length before its ciliated, green tip emerges within summer days elapsed between the appearance of 112

the furled leaf blades within the sheath of the 2O0 terminal leaf, and when they achieved their final 80 Mid I lengths; eight to thirteen days later, the leaf NODE 6° N.Z 3l 4CII ~° '? i ,'~° , sheaths finished elongating. In C. culeou, foliage II il leaf sheaths are longer than the branch internodes r~oO and persist after the leaf blades have fallen. 30 I Branch nodes are rarely visible without special efforts to remove the sheaths. 29 !I "'"--" II,l The third foliage leaf on a new branch frequently is the first to have a blade with a well- ~o f ~' L~, il'J defined petiole. These leaf blades, however, are i only 22 to 47 ~o of the mean lengths of other foliage leaf blades produced by the same plants. Since they are associated with new branches, these small foliage leaf blades are common on f ., fl II young culms. The few such minileaves present on older culms overlap in size the lower end of the ,, ,l, range of leaf lengths. il .16 l Jl Annual leafproduction. In the course of our study, 24 I n it became apparent that the length of successive N I foliage leaf blades on a branch is not random, but 23 I- " 17 ,, iI that long and short blades occur in definite patterns (Fig. 19). From measurements of leaf blades and interblade distances (measured from the ii base of one petiole to the base of the next) on il numerous branches we came to conclusions that were supported by evidence from marked leaves, and which provided us with a method of reading ,o f,,.,, ill iI the history of leaf production on other bamboo branches (Fig. 20): 19 f N,29 I 1. A single season's foliage leaf blades are closely spaced, separated from the previous season's il production by a greater interblade distance established by growth occurring at the time of iI emergence of the first leaf in spring. 2. When two leaves are produced in a season, the i

Fig. 19. Diagram of the foliage pattern on all live branches at iI the upper 19 nodes of a six-year-old culm collected on Cerro Otto on November 13, 1988. Average length of the branches at each node is drawn to scale (horizontal axis), and average 14 f ~''. . I I,[ i leaf blade length and average interblade spacing on those / branches is drawn to a different scale (represented on verti- NODE ~" N • 2 cal axis). Note the consistent patterns of interblade spacings 13 ~ ,, I and relative leaf blade lengths. Leaf blades are clustered into i N~ cohorts. N = number of branches. 0 I00 150 200 113

2-LEAF COHORTS IN EACH OF LAST 2. YEARS I I 4o N= 25 BRANCHES /,~ I 1986-87 1987-88 I I I I I I I Ioo 150 BRANCH LENGTH (MM)

TERMINAL 2- LEAF COHORTS 8° i PRECEDED BY 3-LEAF COHORTS .J N= 39 BRANCHES lP-.-ll I I 1986 - 87 1987-88 125 I I I 1 / I I I I I 17 5 225 BRANCH LENGTH (MM)

,%-LEAF COHORTS IN EACH 80 OF THE LAST 2 YEARS 40,, N; 15 BRANCHES ,q~.//--../I T = 1986 -87 1987-88 I I I I I I I I I I I I 3,~0 25O 3OO BRANCH LENGTH (MM) Fig. 20. The three most frequent patterns of leaf blade length and interblade spacing on branches (from nodes 17, 21 and 24) of the culm of Figure 19. Some branches produced two leaves in each of the previous two years (above); some produced three leaves in each of the two previous years (below), and in a third category a terminal, two-leaf cohort was preceded by a three-leaf cohort (middle). Mean leaf blade length and mean interblade spacing of leaves produced in the last two years are drawn to scale. Branches that produced three leaves in each of the last two seasons averaged twice as long as branches that produced two crops of only two leaves in those seasons.

second leaf blade is an average 18~o longer tion on a branch by counting the number of co- than the first; when three leaves are produced, horts represented by leaf blades and leaf blade the middle one (the second) is the longest (an scars present. The mean annual production at average 24% longer than the first, 14% longer Puerto Blest calculated in this manner (2.64 leaves than the third), and is longer than the blades per live branch) was significantly greater than at of either 2-leaf or l-leaf annual cohorts other sites (Table 4). (Fig. 20). Initial observations indicated that few leaf . In 3-leaf cohorts, the interblade distance is blades are lost in their first year, so leaf produc- greater between the first and second leaf blades tion was also determined, for six years, by annu- produced than between the second and third. ally marking in early spring the new foliage leaf The spacing between leaf blades decreases to- blades on selected nodes at Cerro Otto and at ward the tip of the branch, suggesting that the Llao Llao. The number of unmarked leaf blades rate of growth of the branch is reduced with each spring revealed the previous year's produc- age. tion. For this study we selected nodes in the middle of the foliage bearing section of the culms, Utilizing the information outlined above, we are and did not include culms less than three years of able to calculate the annual foliage leaf produc- age. Data recorded near both Cerro Otto study 114

clumps demonstrated that branches on the exposed hillside around Cerro Otto Clump ~ 1 annually produced more leaves than did those in the forest near Clump #2 (2.06 + 0.04, N = 424 versus 1.35 + 0.06, N = 214; a highly significant difference). At both sites on Cerro Otto, however, leaf production per branch was significantly greater than in the forests of Llao Llao (Table 4). A third method of estimating annual production of leaves per branch is from a count of leaf blades and leaf blade scars on branches of known age. Fifteen such branches, selected from marked culms of Cerro Otto Clump # 2, revealed an average annual production of 3.1 leaves per branch. Annual leaf production calculated in the above manner and from cohort counts is clearly greater than that determined from marked leaves (Table 4). This is explained in part by the fact that the nodes where leaves were marked had many short, less productive branches, whereas the other methods made use of selected long branches. In addition, cohort counts make no allowance for years in which a branch with a live terminal fails to produce a new leaf, and this occurred in about 2 7o of the marked branches.

Standing crop. The standing crop of foliage leaf blades on a culm of C. culeou varies considerably with the season. New leaves are produced primarily from early November to late January, and during this period the standing crop of foliage leaf blades may increase by 157o in forested areas such as Llao Llao, or by as much as 80~o on Cerro Otto. During the summer months, however, almost one-third of the standing crop is lost. A further small loss occurs through the winter. The greatest number of live foliage leaf blades on a branch (exclusive of secondary branches) was 15, observed on a branch at Puerto Blest. The average number of leaf blades on a branch varies with the position of the node from which the branch emerges and with the age of the branch (Figs. 18B and 21). On a two-year-old culm that has finished its first season of leaf production, Fig, 21. Branchesfromculmsofknown age collected in spring branches seldom have more than two or three leaf from Cerro Otto Clump #2. Note annual clusters of leaf blades plus a basal minileaf. A year later most blades. branches will have four or five leaf blades. The 115 number may increase slightly during the third sea- ally a concentration of elongate cilia on either side son of foliage production, but thereafter remains of the attachment of the leaf blade to the sheath. fairly constant as the number of leaf blades lost Another feature common to all our clumps is the is balanced by the annual production. presence of two distinct rows of microscopic, When considering differences in the standing apically-directed hairs parallel to the margin of crop of foliage leaf blades among study sites, we the foliage leaf sheath and proximal to the outer can reduce seasonal variation by counting leaves ligule. before new crops emerge in the spring, and we Because length of foliage leaf blades is used as can reduce variations due to age of the culm and a taxonomic character (Freier 1941), we exam- nodal position of the branches by discarding mea- ined the reliability of this criterion in C. culeou. surements of two-year-old culms and considering Over five years we measured more than 40 only branches in the middle of the branch-bearing samples of leaf blades, including dried blades from section of the culm. This done, there are still sig- under monitored plants and green leaf blades nificant differences in the standing crops of foli- sampled randomly or as nodal complements on age leaf blades of C. culeou clumps at Cerro Otto, culms of known age. Leaf blade length was no- Llao Llao, and at La Veranada (Table 4). tably variable and could be correlated with many factors. In four tests, however, fresh foliage leaf Morphology of foliage leaf blades. Foliage leaf blades did not change lengths when dried. blades of C. culeou are elongate, with a prominent Average lengths of fallen foliage leaf blades central vein and a sharp point at the tip. Leaf from different areas under a clump were statisti- blades of clumps in sheltered sites with low ex- cally different (Table 5, Cerro Otto Clump # 1, posure indices are more lax and softer to the touch 1985). To minimize this effect we mixed handfuls than those growing in exposed areas. Foliage leaf of leaves from eight different areas under a clump blades from clumps at Cerro Otto, Llao Llao and and measured subsamples of 43 to 365 leaf blades. Puerto Blest have many small hairs on the abaxial Samples collected in different years under a spe- surface; those from La Veranada and from cific clump were more similar to each other than Castafio Overo have few or none. The edges of they were to samples from another clump in the the leaf blades tend to be white, somewhat hard, same area, and samples from both clumps in a and are bordered by a series of stiff, apically- specific study site were more similar to each other directed, curved spines (trichomes) that measure than they were to samples from clumps at a dif- as much as a millimeter in length. Leaf blades are ferent site (Table 5). Finally, if the mean of all leaf asymmetrical: the half of the blade that was pe- blade lengths for each monitored clump is plotted ripheral in the furled condition is always some- against ambient light intensity (either index from what wider, often with more secondary veins, and Table 1), a significant negative correlation is evi- with fewer trichomes on the border. dent (Fig. 22). Most leaf blades have two secondary veins on The length of the foliage leaf blade is related to either side of the central vein; in some, there are the length of the foliage leaf sheath from which it two on one side and three on the other. The num- emerges: Length sheath (mm)= 16.49 + 0.155 x ber of tertiary veins between these varies from 24 length blade (mm); N = 71, r = + 0.47. Foliage leaf to 40 (counted in the widest part of the blade). sheath lengths are less variable than blade lengths: Interconnecting the tertiary veins, are small trans- coefficients of variations are 8.5~o and 11.6~o, verse veins (tessellations) that are visible in the respectively. Length of a leaf blade also correlates leaf blades of all our marked clumps, although with the length of the branch it is on and its po- they are more conspicuous in leaf blades of some sition on the branch. Shorter branches (which are plants than in others. also younger branches) have shorter leaf blades; The inner ligule and the distal end of the foli- regressions of the length of the terminal leaf blade age leaf sheath have many cilia, and there is usu- and of the penultimate blade against branch length 116

Table 5. Chusquea culeou: length of foliage leaf blades a.

Clump Date N Mean + se (ram) Mean for clump (mm)

Puerto Blest # 1 11/85 73 90.96 + 2.30 11/88 81 89.43 + 2.41 11/89 100 91.45 _+ 2.14 90.66 + 1.31

Puerto Blest # 2 11/85 43 102.14 + 3.27 11/88 78 103.59 + 2.97 11/89 80 97.87 _+ 2.72 101.00 _+ 1.72

Castafio Overo # 1 11/84 62 40.61 + 1.30 11/85 114 41.51 _+ 1.26 11/88 365 42.35 _+ 0.62 11/89 209 40.33 _+ 0.77 41.52 + 0.43

Castafio Overo # 2 11/85 89 48.33 + 2.26 11/88 144 53.16+ 1.65 11/89 150 51.80 + 1.71 51.50 + 1.05

Llao Llao # 1 11/85 67 80.22 + 2.42 11/88 114 68.70 + 2.01 11/89 117 75.05 + 1.83 73.78 + 1.18 NS

Llao Llao # 2 11/85 54 76.24 + 2.09 11/88 119 70.94 + 2.20 10/89 121 76.72 + 1.95 74.29 _+ 1.26 NS

La Veranada # 1 10/85 117 41.55 + 0.96 11/86 118 34.92 + 0.87 38.22 + 0.65

La Veranada #2 10/85 94 41.37 + 0.97 11/86 86 43.37 + 0.95 11/88 145 38.85 + 0.97 11/88 164 42.27 _+ 0.94 10/89 136 45.74 + 0.93 42.25 + 0.44

La Veranada # 3 10/88 240 48.18 + 0.86 10/89 173 46.48 + 0.87 47.47 + 0.62

Cerro Otto # 1 10/85 86 58.99 + 1.73 11/85A b 82 63.09 + 1.34 B b 97 61.16+ 1.10 C b 78 57.60+ 1.26 10/88 243 53.10 + 0.77 10/89 149 56.98 + 0.80 57.23 + 0.44

Cerro Otto # 2 11/85 126 77.52 + 1.77 10/88 126 66.21 + 1.70 10/89 140 71.93 + 1.70 71.89 + 0.99 a All samples collected from ground under clumps. The range was 20-160 mm. b Duplicate samples from different areas under the clump. NS: This pair not significantly different by T-test. Pairs at all other sites were significantly different. have significant positive slopes (r= +0.61; (e.g., Fig. 13F). To analyze this further, measure- r = + 0.82). ments were made of all leaf blades on samples of The average length of foliage leaf blades from three branches of comparable length collected different nodes on a culm may differ significantly from each node between the 25th and 30th nodes 117

105 • the number of leaf blades on the branch, and the Puerto Blest • Castano 0vero A position of the leaf in the annual sequence of leaf 90. • Lloo Llao • production. Mean blade length also varies with i" ~ La Veranada 0 light intensity, location and with the individual clump. This analysis documents the difficulty of ;~75,~ . ZX 0 60. selecting a representative sample of foliage leaf blades of C. culeou, and suggests that as a taxo- =~ 45 nomic tool length of the foliage leaf blade should be used with caution. 30 t t t t ° t t t ",, 5 10 15 20 25 30 35 40 Light intensity index Survival of foliage leaf blades. Although foliage Fig. 22. Relationship of mean length of foliage leaf blades in leaf blades survived longer in the forest on Cerro monitored bamboo clumps (Table 5) to light intensity index at Otto than in the more exposed site nearby, two- clumps, measured at 30 cm (see Methods; Table 1).r = - 0.86 thirds of the marked blades on Cerro Otto died (Y = 91.4 - 1.55X). and fell before they were two years of age, and none survived six years (Fig. 23). Survival was of three culms. On one of the three culms, mean better in the Llao Llao forest, where 71% of the length of the blades varied significantly with the blades reached two years of age and 12% lived six node number (analysis of variance, p < 0.01). That years or more (Fig. 23). is, each node on a culm tends to produce leaf Information gained from analysis of the spac- blades of a specific length, although mean leaf ing and length of foliage leaf blades on a branch blade lengths at comparable nodes on different (illustrated in Fig. 20) enables us to identify co- culms of the same plant are not the same. The age horts and determine the age of leaf blades on of the culm is partly responsible for this effect, for different plants and at study sites where blades even with minileaves excluded from the calcula- were not marked. Compared with curves derived tions, leaf blades on a two-year-old culm were from marked leaf blades, cohort analysis indi- significantly shorter (p < 0.001) than those from cates better survival at Cerro Otto to two years of comparable nodes on older culms of the same age and a more rapid decline at Llao Llao plant at La Veranada (Nov., 1986): (Fig. 24). Both methods of determining survival, however, indicate that most foliage leaf blades two-year-old culm: N = 55, mean = 37.44 + 0.86 mm live two years or more, and few survive six years. three-year-old culm: N = 89, mean = 48.78 + 1.25 over four-year culm: N = 93, mean = 46.19+ 0.95 The average length of foliage leaf blades on a 100 ~A branch is positively correlated with the number of leaf blades present on that branch (N=35, • f o Llao r= +0.47, p<0.01). By analysis of variance of 60 the same data it was found that the leaf blades on E each branch tend to have a characteristic length Q.~ 40 ~Cerro0tto A,, (p < 0.001). This does not contradict the fact that J . ,, 20 leaf blade length varies along each branch (long- "'A ...... "A short-long, etc.; Fig. 20), for the whole series is 0 : : ; -0~----_~ @ shorter or longer on each branch. 1 2 3 4 5 6 7 Years of age In conclusion, the length of a foliage leaf blade Fig. 23. Survival of marked foliage leaf blades at Cerro Otto in C. culeou is correlated with the age of the culm (2105 blades) and Llao Llao (660 blades) between 1985 and it is on, the position of the node from which it 1991. Blades were marked when approximately one year of emerges, the age and length of the branch it is on, age. 118

lOO ...... Parasites

8o ,\.,""&, ...f Castano Overo (61) The most obvious parasites on Chusquea are larvae of leaf-roller moths (Tortricidae; identifica- 60 ~r~.. - "~ Llao Llao (99) tion by W. W. Middlekauf and J. Powell based on larvae from bamboo shoots and adults col- ne 40 Cerro Otto (113)----.~:~, "X~.." "'.'o "'. lected in bamboo thickets). These larvae attack 2o Po. o B,.t "° Ver°°°d° (1°7) and kill over 40 % of the new shoots. Their presence is indicated by a small hole (1 mm diameter) 0 1 2 3 4 5 6 in the culm sheaths near the top of new shoots. Years of age Removal of the sheaths reveals in the core of the Fig. 24. Survival of foliage leaf blades at five localities by shoot a spiral groove that leads to the larva. A analysis of annual cohorts of leaf blades and blade scars. The shoot usually contains only one moth larva, which number of branches analyzed for each curve is indicated. may be two centimeters or more in length. Addi- tional insects found on dead shoots may be sec- Litter ondary invaders following the moth larvae into the interior of the shoots (including four species The dry weight of the litter produced by indi- of larval Diptera, beetle larvae, and adult short- vidual bamboo clumps was remarkably consis- winged beetles). tent from one year to the next. Annual averages Large white grubs each weighing as much as varied from 56 gm/m 2 for Castafio Overo Clump 4.3gm were sometimes common in the soil # 2 to 460 gm/m 2 for Cerro Otto Clump # 1 around clumps of C. culeou, especially at Castafio (Table 2). The former is a vigorous small clump; Overo. They are larvae of a stag beetle (Lu- the latter is large and declining (see Table 3). The canidae, identified by Dr. Paul Richter as close to average annual litter production of all clumps was Lucanus), but their relationship to the bamboo is 167 gm/m 2. Annual foliage litter in a C. culeou unclear. C. culeou is utilized lightly by a variety thicket in Chile was 385 and 425 gm/m 2 (Veblen of small mammals. We have seen the mouse et al. 1980; Veblen 1982); a comparable figure for Irenomys tarsalis shinny up bamboo culms and an Asiatic bamboo was 300-375 gm/m 2 (Numata climb about in the foliage. Dromiciops australis, a 1965). small marsupial, uses bamboo leaves to construct The average weight of litter produced annually its spherical nest. A variety of small rodents eat by a clump was correlated with the number of the underground rhizome buds and emerging culms/m 2 in that clump (r = + 0.94, N = 8 clumps) shoots of C. culeou, among them Abrothrix lon- but not with the average diameter of the culms. gipilis, Aconaemys fuscus, Irenomys tarsalis (Pear- Average number of foliage leaf blades/m 2 in the son 1983), Auliscomys micropus and Ctenomys litter under a clump was not related to either culm mendocinus. Such predation contributes to a high measurement. mortality of rhizome buds in some areas. Once a Since areas of the monitored clumps and the bamboo shoot has reached 10 cm in height it is no number of live culms within them are known, the longer attractive to small rodents, although ma- mean annual litter production of each clump and ture rhizomes are eaten by the fossorial Ctenomys each live culm within a clump can be calculated mendocinus. (Table 2). Combining data for all clumps, the av- In addition to rodents, young bamboo shoots erage annual litter production per clump weighed are eaten by livestock and (rarely) by humans. 920 gm (DW) and included 41,267 foliage leaf The foliage is browsed by horses, cattle, sheep, blades; the overall average annual production per and by the introduced European hare. The heavi- live culm was 5.1 gm (DW) including 225 foliage est browsing is done by cattle: young culms with leaf blades. their first crop of leaves are browsed at every 119

node, and the tender tops are nipped off and con- for many years and reported that all bamboo on sumed. Heavy browsing produces a bottle-brush the peninsula bloomed in 1939. There has not effect in mature culms, with many short dead been a similar bloom on the popular peninsula branches of equal length at each node. In areas since 1939, which clearly indicates that the flow- with many cattle, the bamboo thrives only where ering cycle in this population is longer than 54 livestock cannot reach it. years. Additional evidence of a long life cycle Branch buds are also susceptible to parasites. comes from the apparent advanced age of our We have found aphids, thrips, white lice-like in- marked clumps on the peninsula, and from the sects, microscopic brown mites and a caterpillar amount of dead material, including shrubs, hard- living in and around young buds. Some insects wood saplings and bamboo culms, within the that pupate within a terminal furled leaf blade dense thickets of bamboo. prevent its development, and mature leaf blades In the Nahuel Huapi area, C. culeou does not sometimes show evidence of insect bites. seem to be committed to gregarious blooming. A few plants bloom and die each year, isolated from other blooming plants. These produce few seeds, Flowering and there is a high probability that these bamboos are self-sterile. We have found flowering clumps The inflorescence of Chusquea is an aggregation in all our study areas. To quantify the amount of of spikelets arranged on a branch in close or open flowering that was occurring, we counted flower- clusters (panicles). The arrangement of the spike- ing plants along 3 km of the unpaved road at lets and the morphology of spikelet and flower Puerto Blest in 11 of the past 13 years. None were components are taxonomically useful (Clark seen in 1980, but in all other years between 5 and 1989). 18 flowering plants were recorded. Reproductive Due to the length of the vegetative phase, there asynchrony reduces the effectiveness of gregari- is no documentation of the length of the life cycle ous flowering as a stratagem for ridding the of C. culeou. Statements concerning intervals of species of its parasites; self-sterility provides a 15 to 25 years between mass flowerings of this mechanism to encourage gregarious blooming. species in southern Chile and Argentina are pri- In some flowering clumps observed by us, all marily derived from earlier tentative statements culms were in flower at the same time; in others, (e.g., Hosseus 1915; Gunckel 1948), which some culms produced flowers one year, and other through repetition have achieved an air of authen- culms flowered in the subsequent year. Also noted ticity. In our own interviews with 14 long-term, were culms that had both dried flowering credible residents of the Nahuel Huapi area, 11 branches from the previous year and new inflo- reported widespread flowering of bamboo about rescences emerging from the same node. We have 1940 in areas from Lago Futalaufquen north to not found new culm shoots emerging under fully Villa Angostura (a distance of 230 km), and in- flowering bamboo clumps, although there were cluding Lago Frias (near Puerto Blest) and Pen- new shoots under a clump in which a quarter of insula Llao Llao. Three of the older informants the live culms flowered one year and subsequently recalled flowerings between 1900 and 1904 in ad- no further flowers appeared. It is possible this dition to the 1940 bloom, although the extent of clump was a merger of plants with different ge- the area involved in the earlier bloom was not netic programming for flowering behavior; a simi- known. There were also reports of areas of a few lar case was described by McClure (1966, hectares with concentrations of blooming plants page 85). in 1950-51, 1968 and 1986. In summary, the length of the vegetative cycle Some evidence on the length of the life cycle of of C. culeou in Parque Nacional Nahuel Huapi C. culeou around Nahuel Huapi comes from Juan remains unknown; evidently in the Llao Llao Chihuay, who lived on the Llao Llao Peninsula population it is longer than 54 years. The flow- 120

ering behavior of this bamboo in the temperate rhizome buds (usually only 2). The first five gen- climate of the park is variable: there has been at erations of tiny rhizomes in this young plant each least one gregarious bloom over a large area, sev- had produced only a single successor rhizome. As eral sporadic blooms over smaller areas, and seedling C. culeou produce a single new rhizome some populations perennially include a few flow- and culm each year in this temperate climate ering plants. In our experience, clumps of C. eu- (Figs 1 and 2 in Haverbeck 1983), the first five leou always die after producing flowers. Simulta- rhizome generations in the young plant from neous death of all the bamboo in the forests must Cerro Otto represent five years of growth. In the have a considerable impact on plant succession. sixth year, two rhizome buds developed from the It opens up the forest to invasion by aggressive fifth year rhizome and produced the first culms plants such as Rosa eglanteria and Berberis, or to over 1 m in height. Analysis of leaf cohorts dis- the reseeding of the Nothofagus trees (Lebedeff closed an age of 8 years on the oldest living culm 1942; Veblen 1982). In addition, since dead culms on this plant, which was culm # 10 in the series remain standing for many years, the forest could and was the product of at least 7 generations of be expected to be especially vulnerable to fire at rhizomes. The plant itself was thus at least 15 this time. years old, and the largest culm still was not as large as culms of nearby plants. In seven years of observations, we consistently Seedlings have found more flowering plants and more seedlings at Puerto Blest than at the other four study Although scattered C. culeou plants in flower can sites, suggesting that either genetic differences be found, they rarely produce seeds. In autumn, among bamboo populations or environmental at a site where several plants had bloomed, only factors affect the flowering behavior. about 2~o of the spikelets contained seeds. The seeds look like small wheat grains and have an average size of about 5.6 × 1.8 mm. The occa- Clump growth and longevity sional discovery of a few seedlings or young clumps is evidence that some fertilization and Although we selected for study bamboo clumps seed production occurs. Seedlings are not found that were discrete enough to suggest that the directly under a presumptive parent plant, but clump was a single plant, it is possible that some there is usually a dead bamboo plant within 10 m of our clumps developed from more than one of a seedling. seed, or that they were a clonal subdivision of a The first young culms on a seedling are slender nearby plant, as occurs in bamboo species with (about 2 mm diameter) and less than 15 cm tall. leptomorph rhizomes. We have not seen lepto- In subsequent years increasingly larger culms are morph rhizomes, however, and C. culeou in the produced until the mature size is attained. One area of our study does not spread into adjacent young plant excavated in its entirety on Cerro open areas. On the Llao Llao Peninsula, many Otto consisted of 19 culms (the smallest ones abandoned paths shaded by a canopy of arching dead) in an evenly graded series ranging from 2 bamboo culms are now blocked by hardwood to 20 mm diameter at ground level and from 17 saplings, but the nearby bamboo has not en- to 430 cm in height. The generations of rhizomes croached upon the paths. About 1983, vegetation terminating in these culms likewise were in a was cleared along the paved road through the graded series, and the gradient of culm diameters peninsula, leaving an open border between the matched well the sequence of their emergence in road and the forest of dense bamboo and large the rhizome series. The smaller rhizomes (less Nothofagus dombeyi trees. In 1990 the road was than 1 cm in length) had short necks of 2 to 5 lined by young beech trees and not by bamboo. segments and a limited number of segments with Over thirteen years we have repeatedly photo- 121 graphed a seldom-used, unsurfaced jeep trail at ited potential cannot compensate for the death of Castafio Overo, looking for changes in the C. cu- the older culms, so this clump is slowly dying. The leou clumps that line the trail. Comparing the reduced precipitation in recent years may be photographs, we find no evidence that the bam- partly responsible. The number of live culms boo in this area has expanded aggressively into present in the other marked clumps increased be- the open areas. tween 1.3 and 16.2~o per year (Table 3), with an The history of land use on the Llao Llao Pen- average annual rate of increase for all clumps insula documents that the present distribution of of[ 4.64~o. The rate of increase was negatively bamboo in this area was determined at the last correlated with clump size (Annual ~o mass flowering (around 1940), rather than by ag- change= 12.3-0.056x number of live culms; gressive vegetative growth. The last cattle were N= 10, r= -0.67, p<0.05) and would slow to removed from the peninsula in 1936, according to zero when clump size reached 220 live culms. informant Juan Chihuay, a resident there at the These data may be used to estimate longevity time. With the flowering following closely on the of a clump. If one starts with the assumption that abandonment of the pasture areas, bamboo suc- a ten-year-old clump has ten living culms more cessfully seeded extensive open areas, resulting in than 1 m in height, a constant rate of increase of the present bamboo thickets on the peninsula. An 4.64~o compounded annually would produce in occasional dead and dried Rosa eglanteria in the 68 years a clump with 138 live culms (approxi- thickets attests to the fact that the area was once mately the average number of culms in the moni- open and grazed. tored clumps). If, however, the rate of increase in Measurements of the areas occupied by each of number of culms is adjusted using the above re- our monitored clumps (Table 2), were repeated in gression to reflect the slower increase in large 1991 and verified the slow expansion of mature clumps, then a bamboo plant with 138 live culms clumps of C. culeou. Pachymorph rhizomes less would be 38 years old. At 65 years it would con- than 15 cm long, and a mean generation time of tain 215 culms, which is approaching the maxi- two or more years between culms, are two factors mum of 220 live culms (when the rate of increase that would contribute to slow expansion of equals the rate of mortality). Clumps larger than clumps. These observations and our years among this fall within the considerable variation inherent the bamboo plants lead us to conclude that in in these calculations, or possibly consist of two or Parque Nacional Nahuel Huapi C. culeou grows more separate plants. primarily where its seeds have fallen. The situa- These calculations suggest that the majority of tion appears to be different in Chile, where our C. culeou clumps are between 35 and 55 years Chusquea culeou is described as an aggressive in- old; the great range in the number of culms per vader, and clumps continuously shift position clump indicates that our widespread samples were through the forest (Veblen 1982). not the result of a single seeding episode. The net change per year of the number of live culms in our monitored clumps was calculated from repeated censuses (Table 3). The prorated Phytomass and net annual production mortality ofculms in each clump, subtracted from the increase in live culms due to successful shoots, From the census conducted through the bamboo provided a measure of the annual change. In one thicket on the Llao Llao Peninsula, the total den- clump (Cerro Otto Clump # 1), the number of sity of culms was calculated to be 187,000/ha, of live culms decreased between censuses, resulting which 36~o were dead (Table 6). Live culms in in a negative net change per year. In the last four the sample had an average diameter of years this large clump had produced a total of five 20.78 + 0.14 mm (N --- 856); dead culms averaged shoots, and thus had only five rhizomes likely to 17.02 + 0.24 mm (N --- 483). Density in a thicket produce another rhizome generation. Such a lim- of Chusquea culeou at a comparable latitude and 122

Table 6. Phytomass and net annual production a.

N Fresh weight Dry weight t/ha t/ha

Phytomass Live culms 120,000 172.3 100.0 1 year old 8,000 (no foliage) 10.0 5.8 > 2 years 112,000 (with foliage) 162.4 94.2 Dead culms 67,000 28.5

Total culms 187,000 128.5 Dead shoots 16,000 1.3 Rhizomes & Roots 82.5 49.5

Total 179.3

Net annual production Litter: 3.7 gm DW litter x 112,000 culms 0.4 Foliage: 118 gm DW on 8,000 2-year-old culms 0.9 Branch growth: 28 mm (0.0108 gm) on 48 x 10 6 branches 0.5 Yearling culms: 723 gm DW x 8,000 yearling culms 5.8 Rhizomes and roots of yearling culms: 39~o Of 723 gm x 8,000 2.3 Dead shoots: 82 gm x 16,000 1.3 Rhizomes and roots of dead shoots: 39~o of 723 gm x 16,000 4.5

Total 15.7 a Based on data from transect on Llao Llao Peninsula and from nearby monitored clumps (Table 2). Ave. mature culm (841 gm DW) minus 14~o foliage = yearling culm (723 gm). Branch increments were weighed; weight of dead shoots estimated from volume calculations. elevation in Chile was 150,000 to 200,000 culms foliage (FW culm without foliage = 10.4 x ra- per hectare (Veblen 1982). dius 2- 20.7), and converted to dry weight. The Using the Llao Llao census data, the regression total dry weight of standing dead culms in the illustrated in Figure 11, and data in Table 6, the thicket was 28.5 t/ha. Combining the dry weights aboveground fresh weight of each live yearling of live and dead culms gives a value of 128.5 t/ha, and live mature culm including foliage was cal- 22 ~o of which was dead culms. This total above- culated and summed, giving an estimate of 172.3 ground mass is somewhat below the average of metric tons per hectare (t/ha) or 100.0 t/ha DW North American and European temperate hard- (DW of culms and foliage is 58~o of FW). This wood forests listed by DeAngelis et al. (1981). phytomass is somewhat less than the 156-162 t/ Based on the weights of excavated rhizomes ha aboveground dry weight reported for a thicket and roots, reported above, we estimate that the of C. culeou in Chile (Veblen et al. 1980). The two underground mass of roots and rhizomes in the thickets consisted almost entirely of bamboo and Llao Llao thicket adds another 82.5 t/ha (FW) or the culm densities were about equal; the differ- 49.5 t/ha (DW) and brings the total dry weight of ence in phytomass is probably explained by the bamboo (live culms and foliage, dead culms, plus larger size of the Chilean culms. underground mass) to 179.3 t/ha, about 28~o of The aboveground phytomass is increased ap- which is underground (Table 6). The total is preciably if dead culms are included, since they slightly less than the average for North American remain standing and accumulate for many years. and European temperate hardwood forests listed The mass of standing dead culms was calculated by DeAngelis et al. (1981). from the regression for live culms minus their The net annual production of the Llao Llao 123 bamboo thicket can be estimated by combining strate an alternating dominance of forest and the census data from the transect through the steppe species in the park in postglacial times. bamboo thicket with growth or production data The change from lateglacial to postglacial climate from the two monitored nearby clumps (data in was most pronounced, but notable changes also Table 2) and some further measurements made occurred about 8500, 6000, 5000 and 3000 years by us. The estimate (DW) is the sum of the fol- ago (Markgraf 1983). Glaciers may have been lowing compartments: litter produced (14 × 10 6 agents in the distribution of bamboo plants, and foliage leaf blades plus branches and sheaths), the the local presence of bamboo would have been foliage produced by 2-year-old culms, the annual modified further by the massive ashfalls common growth increment of branches, the new culms and in this region, and by wildfires. Mature bamboo their rhizomes, and the shoots produced that died plants are able to survive fires that kill beech trees and the rhizomes associated with them (Table 6). (Hosseus 1915), which may explain the presence No compartment is needed for flowers and fruits, of pure stands of C. culeou in some areas. How- since none were produced, and loss ofphytomass ever, the impact of a fire surely depends on tim- to herbivores was trivial. The total net annual ing. A fire sweeping through postflowering dead production above and below ground was 15.7 t/ culms when seedlings are just getting established ha (DW), which is about one-third more than the might be much more destructive than a fire after average dry weight for North American and Eu- the development of large rhizomes. ropean temperate hardwood forest (DeAngelis Although plants may be relocated by storms et al. 1981). Note that about 43~o of the produc- and landslides, C. culeou, with its pachymorph tion occurred underground. rhizomes, does not aggressively invade new areas, Considering only the aboveground portion of and except for periods of gregarious blooming, the bamboo productivity, the net primary annual seed production is unusually low. The seeds production at Llao Llao was 8.9t/ha (DW), themselves are too heavy to be spread far by wind, slightly less than the dry weight production of the even the strong winds common in Patagonia, and same species in the milder climate of Chile (10- the large areas of suitable habitat which lack bam- 11 t/ha; Veblen et al. 1980), and about the same boo (such as on the east side of Lago Gutierrez as the dry weight production in a variety of tem- and in the Nothofagus pumilio forest of Arroyo perate zone hardwood forests (approximately Challhuaco) suggest that birds are not major dis- 10 t/ha, Art and Marks 197l; 9 t/ha, DeAngelis persers. The small home ranges of rodents that etal. 1981). It is clear that even when it is only might prey on bamboo seeds preclude a major one of the understory components in the south- role for them in seed dispersal. Two native deer ern beech forests, bamboo is contributing an im- known to have lived in Parque Nacional Nahuel portant share of the total annual production of Huapi but now quite rare, huemuls (Hippocamelus biomass. bisulcus) and pudus (Pudu pudu), may disperse Our calculations do not include dead culms seeds of Chusquea. Cattle, horses, and the intro- lying on the ground; these were not caught in our duced European red deer (Cervus elaphus) are litter traps and were not counted in the census, probably seed predators also, but because of their but would increase estimates of litter. recent arrival, they probably have not affected the present distribution of C. culeou. Thus the dispersal mechanisms of this bamboo remain un- Discussion known, and the large discontinuities in its distribution suggest they are relatively ineffective. The present distribution of C. culeou in Parque The mountainous nature of Parque Nacional Nacional Nahuel Huapi has been dictated by Nahuel Huapi and the abrupt east-west decrease lateglacial changes and by changes during the last in precipitation (3600 mm in 60 km) combine to 13000 postglacial years. Pollen profiles demon- produce numerous and diverse patches of habitat 124

suitable for bamboo. Such a range of habitats carpy were the evolutionary result of rapid clump might be expected to lead to the evolution of a growth and a need to saturate seed predators; number of bamboo species. A single species, they view synchronous blooming as a further ad- however, has adapted to the diverse environments aptation found in areas where the climate is rela- through phenotypic variability. In the southern tively uniform and the plants are closely spaced, Andean forests in general, there is a lack of spe- so that seed predators could be expected to de- ciation among vertebrates, for a high percentage stroy seeds from more than one plant. These con- of the frogs, birds and mammals belong to mo- ditions seem to be met by Chusquea in Parque notypic genera (Patterson 1992; Vuilleumier 1968, Nacional Nahuel Huapi. Our observations sug- 1985). Perhaps the real question is why there are gest, however, that the synchronous blooming and ten species of Chusquea in Chile (Parodi 1945), subsequent death of the bamboo is a stratagem to where there seem to be fewer barriers to gene escape the parasites that attack the most vulner- flow, and where habitats are less diverse than on able stage of the bamboo life cycle - the shoot. the eastern Andean slopes. In the Pleistocene Moth larvae kill more than 40~o of the new there were possibly more opportunities for the shoots. The gregarious flowering and death for fragmentation of populations and for speciation which bamboos are so famous may have evolved to occur. Climatic changes during that period may to enable the population to escape this tax for the have been more extreme on western slopes, as ten or more years required for new generations of they have been in postglacial times (Markgraf shoots to reach a diameter large enough to sup- 1983). The long life cycles of Chusquea must be port the parasitic larvae. If parasites drive the life part of any explanation of the speciation of this cycle, the interval between flowering episodes plant. would be determined by the number of years nec- Bamboo plays its most spectacular role follow- essary for the parasite population to recover to an ing its rare synchronous blooming, for seed pro- unacceptable density. duction is followed by a spectacular increase of The net annual primary production of a thicket mice. All accounts of this phenomenon in south- of Chusquea approaches that of a temperate hard- ern Argentina agree on the great number of mice, wood forest, yet little of the standing crop and but none document what species contributed to little of the annual production is consumed by the outbreaks. Following the blooming and seed herbivores. Fruits or seeds are produced only at production, the bamboo died over large but un- intervals of several decades, and then only at the documented areas and was replaced promptly by cost of the life of the plant. Large and middle- seedlings. Wherever bamboo grows, one encoun- sized vertebrate herbivores are notably scarce in ters similar accounts of mass blooming followed Nothofagus forests (Pearson & Pearson 1982). by outbreaks of rodents. This has led to specu- Small vertebrates are diverse and abundant but lation about the evolutionary forces that drive are largely dependent on secondary producers bamboo to bloom, set seed, and die synchro- such as fungi and detritus-feeding invertebrates. nously at intervals measured in decades. Janzen A study of habitat association patterns of birds in (1976) hypothesized that synchronous seeding is 12 different habitats, including forests close to our a stratagem by which bamboo saturates its seed bamboo study sites, showed that birds are more predators, thereby ensuring that some seeds will diverse and more abundant in shrub communities survive to germinate. It is not clear, however, why than in the structurally more complex beech for- the bamboo cycle needs to be half a century long ests (Ralph 1985). The Nothofagus dombeyi forest when the life expectancy of seed predators is only with dense bamboo understory on the Llao Llao one or two years if they are mice, or 5 to 10 years Peninsula had a large foliage height diversity and if they are birds. From observations of Indian the largest leaf surface area (cm2/cm3) of any of bamboo species, Gadgil and Prasad (1984) con- the bird habitats, yet had no greater numbers or cluded that a long vegetative period and mono- diversity of birds than simpler forests and had 125

fewer birds than scrub communities. Thus the P. Ralph, Paul Richter and David L. Wagner for large net annual production of bamboo in a beech assistance with the identification of insects, and forest is not correlated with a rich avifauna. Only to Dr. Elisa Nicora de Panza for identification of two species of birds were positively associated some bamboo flowers. We thank Karen Klitz for with bamboo: a hummingbird (Sephanoides the preparation of Figure 1. A special acknowl- sephaniodes) and the chucao (Scelorchilus rube- edgement goes to Dr. Lynn Clark for identifica- cula), a ground-living species that scratches in the tion of the bamboo specimens, for her encour- litter for invertebrates (Ralph 1985). Ralph noted agement, and for her advice in the preparation of that insects seemed to make little use of bamboo. the manuscript. Therefore, a large biomass of bamboo might ac- We are grateful to the following people of Bar- tually depress food densities for foraging birds iloche and nearby areas for sharing their infor- and perhaps for other animals as well. mation about the flowering of Chusquea: Antonio The path of energy through a bamboo thicket, and Flora Albisu de Mayorga of E1 Hoyo, Juan as measured by annual increase in biomass, de- Chihuay, Alfonso Huenchupan, Andr6s and Ellen parts in some respects from the usual partition- Lamuniere, the late Don Diego Nell, Nellie Frey ing of energy in a forest. Not only is there no de Neumeyer, Sigfrido Rubulis, Mr. and production of fruit or seed and no increase in the Mrs. Strukely, Ricardo Vallmitjana, and the late number of plants, but 88~o of the net annual pro- Dr. Rodolfo Venzano. ductivity is channeled into increasing by 7 Yo the number of full-sized stems rather than into in- References creasing the size of existing stems. Less than half the shoots produced survive; in some years only Art, H. W. & Marks, P. L. 1971. A summary table of biom- 11 ~o survive. The production of these doomed ass and net annual primary production in forest ecosystems shoots and their rhizomes (36~o of the total net of the world. In: Forest Biomass Studies, Sec. 25: Yield and production) largely seems to be wasted. Growth, 15th Int. Congr., Union Forest Res. Organ., Misc. Publ. 132, Life Sciences and Agriculture Experiment Sta- tion, Univ. Maine, Orono, pp. 3-32. Boelcke, O. 1957. Communidades herb~tceas del Norte de Acknowledgements Patagonia y sus relaciones con la ganaderla. Rev. Invest. Agric. (B. Aires) 11 (l): 5-98 + 17 plates. Clark, L.G. 1985. Three new species of Chusquea (Gra- We are indebted to the Administraci6n de Parque mineae: Bambusoideae). Ann. Misso. Bot. Gard. 72: 864- Nacional Nahuel Huapi for graciously allowing 873. us to pursue our investigations. We also wish to Clark, L. G. 1989. Systematics of Chusquea section Swalle- thank the many parkguards who shared their nochloa, section Verticillatae, section Serpentes, and sec- knowledge and observations with us. Carlos tion Longifoliae (Poaceae-Bambusoideae). Syst. Bot. Monogr., 27: 1-127. Martin and Monica Mermoz of the Grupo de DeAngelis, D. L., Gardner, R. H. & Shugart, H. H. 1981. Investigaci6n Ecol6gica of Parque Nacional Productivity of forest ecosystems studied during the IBP: Nahuel Huapi were most helpful in providing us the woodlands data set. In: Dynamic properties of forest with information and useful discussions. We ac- ecosystems. (D. E. Reichle, ed.). Int. Biol. Programme 23, knowledge with pleasure the hospitality, compan- pp. 567-672. Cambridge Univ. Press, Cambridge. Freier, F. 1941. Contribuci6n al estudio de la anatomia foliar ionship and assistance of Abel Basti and Miguel de las especies del g6nero 'Chusquea' de la flora Argentina. Pellerano, who were in service with Parques Na- Rev. Argent. Agron. 8: 364-379. cionales when we started collecting information Gadgil, M. & Prasad, S. N. 1984. Ecological determinants of on the bamboo. In addition, Michael Christie, life history evolution of two Indian bamboo species. Bio- Rosendo Fraga, Gilberto Gallopin, Sandia Ivey, tropica 16: 161-172. Peter Pearson, Richard Sage and Alberto Sosa Gallopin, G.C. 1978. Estudio ecoldgico integrado de la cuenca del Rio Manso Superior (Rio Negro, Argentina). I. gave much appreciated advice and aid in the field. Descripcidn general de la cuenca. An. Parques Nac. We are grateful to Drs. W. W. Middlekauff, Carol (B. Aires) 14: 161-230. 126

Garcia, D. R., Sourouille, A., Gallopin, G. C. & Montafia, C. Parodi, L. R. 1945. Sinopsis de las gramineas Chilenas del 1978. Estudio ecol6gico integrado de la cuenca del Rio genero Chusquea. Rev. Univ. (Univ. Catolica Chile) 30: Manso Superior (Rio Negro, Argentina). II. Tipos de veg- 61-71. etaci6n. An. Parques Nac. (B. Aires) 14: 231-248. Patterson, B.D. 1992. A new genus and species of long- Gunckel L., H. 1948. La floraci6n de la quila y del colihue en clawed mouse (Rodentia: Muridae) from temperate rain- la Arancania. Cien. Invest. (B. Aires) 4: 91-95. forests of Chile. Zool. J. Linn. Soc. 106: 127-145. Haverbeck M., R. F. 1983. Estudio del crecimiento, variaci6n Pearson, O. P. 1983. Characteristics of a mammalian fauna morfol6gica y reacci6n al corte de coligue (Chusquea culeou from forests in Patagonia, southern Argentina. J. Mammal. Desv.) en un bosque de coigue-tepa-mafiio, en el predio San 64: 476-492. Pablo de Tregua, Panguipulli. Thesis, Univ. Austral, Pearson, O. P. & Pearson, A. K. 1982. Ecology and bioge- Valdivia, Chile. ography of the southern rainforests of Argentina. In: Mam- Hosseus, C. K. 1915. Las cafias de bamb6 en las cordilleras malian Biology in South America. (M. A. Mares & Geno- del sud. Bol. Minist. Agric. (Rep. Argent.) 19: 195-208. ways, H.H., eds.). 6: 129-142. Spec. Publ. Ser. Janzen, D. H. 1976. Why bamboos wait so long to flower. Pymatuning Lab. Ecol., Univ. Pittsburgh, Pennsylvania. Ann. Rev. Ecol. Syst. 7: 347-391. Porterfield, W.M. 1933. The periodicity of bamboo culm Lebedeff, N. 1942. lnforme preliminar sobre los estudios de structures. China J. 18: 357-371. los bosques en la reserva 'Lanin'. Bol. Forestal correspon- Ralph, C. J. 1985. Habitat association patterns of forest and diente a los afios 1938-1939-1940. Dir. Parques Nac., steppe birds of northern Patagonia, Argentina. Condor 87: B. Aires, 221 pp. 471-483. Markgraf, V. 1983. Late and postglacial vegetational and cli- Veblen, T. T. 1982. Growth patterns of Chusquea bamboos in matic changes in subantarctic, temperate, and arid envi- the understory of Chilean Nothofagus forests and their in- ronments in Argentina. Palynology 7: 43-70. fluence in forest dynamics. Bull. Torrey Bot. Club 109: McClure, F.A. 1966. The Bamboos. A Fresh Perspective. 474-487. Harvard Univ. Press, Cambridge, Mass. xv + 347 pp. Veblen, T. T., Schlegel, F. M. & Escobar R., B. 1980. Dry- Mermoz, M. & Martin, C. 1987. Mapa de vegetaci6n del matter production of two species of bamboo (Chusquea Parque y La Reserva Nacional Nahuel Huapi. Secretaria culeou and C. tenuiflora) in south-central Chile. J. Ecol. 68: de Cienc. y Tec. de la Nacion, Delegacion Regional Pat- 397-404. agonia, Argent. 22 pp. Vuilleumier, F. 1968. Origin of frogs of Patagonian forests. Nicora, E. G. 1978. Flora Patag6nica, Parte III Gramineae. Nature 219: 87-89. Colecc. Cientifica INTA, Tomo VII1, (M. N. Correa, ed.), Vuilleumier, F. 1985. Forest birds of Patagonia: ecological B. Aires. geography, speciation, endemism, and faunal history. In: Numata, M. 1965. Ecology of bamboo forests in Japan. In: Neotropical ornithology. (P. A. Buckley, M.S. Foster, Adv. Front. P1. Sci. 10: 89-119, New Delhi. E. S. Morton, R. S. Ridgley & F. G. Buckley, eds.). Ornith. Parodi, L.R. 1941. Estudio preliminar sobre el g6nero Monogr. 36: 255-288. Am. Ornith. Union, Wash. 'Chusquea' en La Argentina. Rev. Argent. Agron. 8: 331- Ward, R. T. 1965. Beech (Nothofagus) forests in the Andes 345. of southwestern Argentina. Am. Midl. Nat. 74: 50-56.

Biology of the Bamboo &lt;Emphasis Type="Italic"&gt;Chusquea Culeou &lt;/Emphasis&gt; (Poaceae: Bambusoideae) in Sout

Biology of the Bamboo <Emphasis Type="Italic">Chusquea Culeou </Emphasis> (Poaceae: Bambusoideae) in Sout