Journal of Vegetation Science 18: 371-378, 2007 © IAVS; Opulus Press Uppsala.

- Interactive effects of introduced herbivores and post-flowering die-off of bamboos -

371

Interactive effects of introduced herbivores and post-flowering die-off of bamboos in Patagonian Nothofagus forests Raffaele, Estela1*; Kitzberger, Thomas1 & Veblen, Thomas2 1Laboratorio

Ecotono-CRUB, Universidad Nacional del Comahue-CONICET, Quintral 1250, (8400) Bariloche, Pcia de Río Negro, Argentina; E-mail [email protected]; 2Department of Geography, University of Colorado, Boulder, CO 80309, USA; E-mail [email protected]; * Corresponding author; Fax + 54(2944)422111; E-mail [email protected]

Abstract Question: In November 2000, Chusquea culeou, a bamboo species dominating Andean forest understories in southern Argentina and Chile, massively flowered and died over a north-south distance of ca. 120 km. Because bamboo is the major forage for large herbivores in these forests, we examined the interactive influences of the bamboo die-off and herbivory by introduced cattle on understory and tree regeneration. Location: Lanín National Park, Argentina. Methods: Permanent plots, in and outside livestock exclosures, were installed in a Nothofagus dombeyi forest in patches of flowered and non-flowered C. culeou. Plots were monitored over four years for changes in understory composition and tree seedling densities and heights. Results: After the C. culeou die-off, new establishment of N. dombeyi was low, both with and without herbivory. Livestock alone directly increased N. dombeyi seedling mortality through physical damage. However, tree seedling browse ratings and height growth were interactively affected by bamboo flowering and herbivory; unfenced plots in flowered bamboo patches had the shortest seedlings, highest browse ratings, and lowest tree seedling annual growth rates. Understory cover was higher where livestock were excluded, and this effect was intensified in the patches of flowered bamboo. Neither herbivory nor bamboo flowering resulted in major changes in species composition, with the exception of Alstroemeria aurea. Conclusion: Effects of livestock on N. dombeyi regeneration were contingent on flowering of C. culeou. Prior to introduction of livestock, N. dombeyi regeneration was probably successful beneath canopy gaps during windows of opportunity following bamboo die-off, but now livestock impede tree regeneration. Herbivory during bamboo withering periods also produces more open understories, particularly affecting palatable heliophyllous herb species such as Alstroemeria aurea. The results underscore the importance of assessing herbivore impacts on tree regeneration during relatively short periods of potential tree regeneration immediately following rare bamboo flowering and die-off. Keywords: Bamboo; Livestock effect; Nothofagus dombeyi, Mast flowering, Seedling establishment. Nomenclature: Correa (1969-1984). Abbreviations: L = Livestock; F = Bamboo flowering; T = Time.

Introduction Chusquea bamboos in the temperate forests of southern Chile and Argentina are keystone species in the sense that they dominate forest understories and impede tree regeneration during their long vegetative intermast periods (Veblen 1982; Pearson et al. 1994; González et al. 2002). At intervals of 30-70 years (depending on the species), however, most species of Chusquea bamboos synchronously flower, seed and subsequently die-off, inducing profound changes in abiotic and biotic conditions in forest understories over large areas (e.g. hundreds of thousands of hectares). Massive die-off of Chusquea bamboos increases the availability of solar radiation in the understory and probably also induces sharp changes in availability of soil nutrients and moisture (Veblen 1982; Pearson et al. 1994; González et al. 2002). Massive bamboo flowering produces large quantities of dry fine fuels, which alters the risk of widespread, severe fires (Veblen et al. 2003). Synchronous flowering and production of enormous numbers of bamboo seeds results in a dramatic increase in resources for small granivores (Pearson et al. 1994) while the death of bamboo culms is a major loss of forage for large herbivores (Veblen 1982; Veblen et al. 1992). Irruption of rodent populations in response to bamboo flowering events and the subsequent crashes of their populations have long been the subjects of substantial ecological interest (Janzen 1976; Jaksic & Lima 2003). The effects of flowering events of Chusquea quila and C. montana on tree regeneration in lower montane and upper subalpine forests, respectively, have been investigated in the southern Andes (González et al. 2002; Holz & Veblen 2006). However, the interplay of bamboo flowering with large introduced herbivores in affecting tree regeneration processes and understory community dynamics has not previously been addressed. Periodic massive flowering and die-off of Chusquea bamboos have long affected the dynamics of Nothofagus forests in the southern Andes of Argentina and Chile, but the ecological consequences of such events have probably

372

Raffaele, E. et al.

been significantly altered by the introduction of livestock to these forests. Large native herbivores such as the guanaco (Lama guanicoe) and the huemul deer (Hippocamelus bisulcus) are not believed to have been abundant in the mesic forests where Chusquea bamboos dominate understories and impede tree regeneration. Instead, these native herbivores favored drier, more open habitats such as the Patagonian steppe and ecotones between the steppe and xeric woodlands (Veblen et al. 1996). Livestock are recent introductions, and have only become abundant in South American temperate forests during the past century and a half (Veblen et al. 1996). Cattle, in particular, are abundant in many areas of Nothofagus forests with bamboo understories at mid-latitudes (ca. 37-43° S). Locally, introduced red deer (Cervus elaphus) may also be abundant (Veblen et al. 1989). While the role of these herbivores during vegetative states of bamboo has been previously assessed in southwestern Argentina (Martín et al. 1985; Veblen et al. 1987, 1992; Vázquez 2002), the interaction of livestock herbivory with bamboo flowering as compounded influences on forest dynamics has not been investigated in the southern Andes, nor to our knowledge elsewhere. Considering that bamboo foliage is the main food item of cattle in species-poor understories of Nothofagus forests (Martín et al. 1985; Veblen et al. 1992; Relva & Caldiz 1998; Vázquez 2002), temporary disappearance of forage during die-off may induce major changes in cattle foraging patterns, including a shift to normally less preferred species. If seedlings of both Chusquea bamboos and of the dominant tree species depend on periods of increased resource availability (especially solar radiation) following massive bamboo flowering and die-off, then the effects of herbivory by livestock during these windows of opportunity are critical to understanding forest dynamics. In November 2000 a massive flowering of the bamboo Chusquea culeou occurred in the Andes of southern Argentina and adjacent parts of Chile at ca. 40° S. This is the only common Chusquea species in southern Argentina. It forms dense thickets with tall shrubs and shrubby trees (e.g. Nothofagus antarctica) and also dominates understories of tall Nothofagus forests from 35 to 47° S (Pearson et al. 1994). The 2000 flowering event affected most of the Andean forests over a 120 km north-south extent centered on ca. 40 °S. This is the first documented massive flowering of this species, anywhere in Argentina, since 1940 when the population synchronously flowered over a 230 km north-south sector of the Andes (Pearson et al. 1994). Unlike other bamboo flowering events such as those of C. quila and C. montana (Gonzalez et al. 2002 and Holz & Veblen 2006), this event left non-flowered patches (from < 1 to several ha in size) surrounded by patches of flowered bamboo. The 2000 flowering event

provided a rare opportunity to examine the interactive effects of the reproductive cycle of C. culeou and introduced herbivores on tree regeneration and understory composition. Study area The study area is located at an elevation of ca. 900 to 950 m near the northwestern end of Lago Falkner (40°27' S, 71°31' W) in Lanín National Park, Argentina. Lago Falkner is one of many glacial lakes in the pre-cordillera physiographic zone separating the main Andean cordillera and the Patagonian plains to the east. Overlying the glacial topography are numerous layers of volcanic ash from which soils are derived. At this latitude, mean annual precipitation decreases sharply from west to east from well over 3500 mm in the Andes near the border with Chile to near 800 mm at the ecotone of woodlands with the Patagonian steppe. Precipitation occurs mainly during April through September (mostly as snow), and December through February are typically dry (Kitzberger & Veblen 2003). The site sampled is a continuous mesic forest of Nothofagus dombeyi, a tall evergreen angiosperm, representing a typical mature forest that developed after severe fires (Veblen et al. 1996). This site is also typical of forest conditions and livestock impacts over extensive areas of the N. dombeyi forest type at latitudes 38 to 44° S in northern Patagonia (Veblen et al. 1992). Along the west-to-east precipitation gradient N. dombeyi forms monotypic stands where the mean annual precipitation is ca. 2000 mm. Further west it dominates stands with coniferous and angiosperm rain forest species, and to the east it dominates stands with the xeric conifer Austrocedrus chilensis. The stand sampled was dominated by trees mostly 100 to 150 years old. Such stands typically develop after severe fires through stages of relatively closed-canopy, even-aged populations to older stands in which treefall gaps provide regeneration opportunities for this relatively shade-intolerant tree species (Veblen et al. 1996). Canopy gaps and presence of large (> 1 m diameter at breast height) fallen trees indicated that this post-fire stand had entered a phase where regeneration of N. dombeyi would be expected to occur beneath canopy openings. The site selected for study has understory vegetation typical of N. dombeyi forests heavily affected by livestock and was being utilized by livestock when we initiated our study in January 2002. We know that this stand had been subject to heavy use by livestock for at least 20 years based on our previous visits and vegetation sampling in the same area (Veblen et al. 1992). The understory in the sampled area is dominated by C. culeou up to 5 m tall and with an average cover of ca. 20%. Elsewhere, in the

- Interactive effects of introduced herbivores and post-flowering die-off of bamboos absence of livestock the understory cover by this species often is 75 to 100%. The shrub stratum is dominated by Berberis darwini, B. linearifolia and Azara lanceolata, and the dominant herbaceous species are Alstroemeria aurea, Osmorhiza chilensis and Adenocaulon chilensis, and the dominant climber is Vicia nigricans. Methods Patchiness in the flowering of C. culeou and cattle presence at the Lago Falkner study site allowed us to develop a factorial research design in which we fenced (i.e. excluded cattle) adjacent flowered and non-flowered bamboo patches to distinguish the effects of livestock and flowering (and their interactions) on tree regeneration and understory vegetation. We monitored vegetation changes over a four-year period in fenced and unfenced plots that we installed in January 2002. This was the first growing season after the flowering in November 2000 and the withering of the bamboo during 2001. Plots were located in interspersed patches of ca. 1 ha of either flowered or non-flowered bamboo. We installed sixteen 15 m × 15 m plots beneath moderately large (ca. 250 m2) tree fall gaps that represent the canopy conditions under which N. dombeyi, the dominant tree species, could potentially regenerate if not impeded by a dense cover of bamboo or other understory species (Veblen 1989). Eight plots (four with live and four with dead bamboo) were fenced, using a 3-m buffer around the 15 m × 15 m plot to exclude cattle. The remaining eight plots were located in areas used by cattle, four in patches with live bamboo (i.e. non-flowered) and four in patches of dead bamboo (i.e. patches flowered in year 2000). Thus, the following four treatments were distributed haphazardly in the interspersed patches of flowered and non-flowered bamboo: fenced- flowered, unfenced-flowered, fenced-non-flowered and unfencednon-flowered. All plots were located beneath canopy openings that appeared to be similar in size and at least five years old. Similarity of size of canopy openings was checked by taking a vertical photograph with a hemispherical lens from the center of each plot at 2 m height to estimate the amount of blockage of potential diffuse and direct sunlight following the procedures of Anderson (1964). Each fenced and unfenced plot included ten systematically located 2-m2 circular permanent microplots. During the austral summers of 2002, 2003, 2004 and 2005, in each microplot we recorded: (1) the percentage cover of all vascular plant species using the BraunBlanquet visual cover scale, (2) numbers and maximum heights of tree seedlings (stems < 140 cm tall); and (3) numbers and heights (vertical distance above the

373

ground) of bamboo pre-existing culms and new (i.e. post-flowering) seedlings. For both tree species and bamboo species we counted first-year (i.e. cotyledon stage) seedlings separately from seedlings that were > 1 year old. We visually assigned browse ratings in each microplot for C. culeou and for each tree species using classes of heavy (3), moderate (2), light (1), or none (0) (Veblen et al. 1992). In each 15 m × 15 m plot, we permanently marked ≥ 30 tree seedlings for subsequent re-measurement of height growth and determination of mortality. Overall, the research design follows a 2 × 2 factorial design in which bamboo flowering (F) flowered and dead versus non-flowered and alive patches, and presence or absence of livestock (L) are the two factors. The combined effects of treatment (fenced and unfenced) and time (T) on dependent variables were tested using repeated measures ANOVA (Gurevitch & Chester 1986). Results Hemispherical photography Mean values for diffuse light factors above the understory did not differ significantly for plots in patches of flowered and non-flowered bamboo nor between fenced and unfenced plots (P > 0.05, Mann-Whitney Rank Sum Test). Thus, differences in the understory vegetation or tree regeneration cannot be attributed to major differences in understory light conditions. Bamboo flowering and livestock effects on tree regeneration of Nothofagus Following the flowering of the bamboo, the rate of establishment of seedlings of N. dombeyi was extremely low and was similar to the rate in patches of non-flowered bamboo. Density of N. dombeyi seedlings bearing cotyledons (i.e. 1-year seedlings) was 0.12 ± 0.3 and 0.15 ± 0.3 (mean ± SE) seedlings/m2 in bamboo-flowered and nonflowered plots, respectively. Nor were there differences in densities of newly established N. dombeyi seedlings between fenced and unfenced treatments (0.15 ± 0.3 and 0.10 ± 0.3 per m2, respectively). In contrast, previously established seedlings (seedlings established before the flowering event) were denser in bamboo-flowered than in non-flowered treatments (F: F1,12 = 5.13, P < 0.05; Fig. 1a). When analysed for change over time, N. dombeyi seedling density declined in unfenced treatments but remained relatively stable in fenced treatments (L × T: F3,36 = 4,84, P < 0.01, Fig. 1a). These declines are consistent with mortality rates estimated from tagged seedlings. Annual loss rates of previously established

374

Raffaele, E. et al. increase in browsing rating between unfenced and fenced plots in flowered patches, but only a 7-fold increase in non-flowered patches. Changes in tagged seedling height over the 2002-2005 period were significantly affected by fencing (L × T: F3,33 = 6.45, P < 0.005), bamboo flowering (F × T: F3,33 = 3.32, P < 0.05) , and their interaction (L × F × T: F3,33 = 3.54, P < 0.05), with differences in height between unfenced and fenced treatments becoming stronger over time in the flowered treatment compared to the non-flowered. By the end of the experiment N. dombeyi seedlings in fenced-flowered plots were ca. 130% taller than fenced seedlings in the non-flowered and 143% taller than unfenced seedlings in the flowered bamboo plots (Fig. 1e). Bamboo flowering and livestock effects on bamboo dynamics

Fig. 1. (a) Density of pre-established (non-cotyledon) Nothofagus dombeyi seedlings, (b) annual N. dombeyi seedling mortality (all causes except desiccation), (c) annual N. dombeyi seedling height growth, (d) mean N. dombeyi seedling browsing ratings and (e) N. dombeyi seedling height over the study period.

seedlings during the four-year period were only affected by differences in herbivory (trampling, browsing) between fenced and unfenced plots (L: F1,12 = 18.88, P < 0.001) and were unrelated to bamboo flowering (F: F1, 12 = 2.09, P = 0.173; Fig. 1b). The interaction of browsing and bamboo flowering had strong effects on annual relative height growth of N. dombeyi seedlings (L × F: F1,12 = 11.97, P < 0.005), with the difference between fenced and unfenced plots being significant only in flowered treatments (Fig. 1c). In flowered and fenced plots, N. dombeyi seedlings grew annually ca. 40% of their height, whereas seedlings in flowered unfenced plots showed on average decreases in height (ca. 5 %/year). Growth rates in the non-flowered treatment were not significantly different between fenced and unfenced plots (P = 0.662, Tukey HSD Test). Likewise, mean browsing ratings measured on tagged N. dombeyi seedlings in 2005 were also dependent on the interaction between fencing and bamboo flowering (L × F: F1,12 = 11.35, P = 0.006, Fig. 1d) , with a 44-fold

New seedlings of bamboo established in higher densities in the bamboo-flowered patches than in the non-flowered patches (F: Fl,12 = 16.46 P < 0.001), but these differences decreased over time after an initial large increase (F × T: F3,36 = 13.56, P < 0.001; Fig. 2a). Averaged over the four years, flowered patches had a density of ca. 49.6 seedlings m-2 (ca. 500 000 seedlings/ha) whereas neighboring non-flowered patches had only minimal bamboo seedling densities (2.25 seedlings/m2; Fig. 2a). Bamboo seedling density increased substantially in year 2003 due to the addition of new one-year seedlings. In years 2004 and 2005, new seedlings emerged at lower rates than 2003, indicating that dormant seeds produced during the year 2000 event germinated from the soil bank over several years, partially compensating for seedling mortality losses. Differences in bamboo seedling density between fenced and unfenced plots were not statistically significant (L: F1,12 = 1.17, P > 0.05) and this result was consistent through time (L × T: F3,36 = 1.51, P > 0.05; Fig. 2a). Withering of the flowering cohort was evidenced by a steady decline in maximum height of (dead) culms in flowered plots from >2 m in 2002 to ca. 0.75 m in 2005 (Fig. 2b). This occurred in strong contrast with non-flowered fenced plots, where live bamboo culms oscillated between 2 and 3 m in height throughout the study period (L × F × T = F3,36 = 4.58, P < 0.05). In the case of non-flowered unfenced plots, live bamboo culms declined in height (Fig. 2b) possibly due to increased browsing pressure. Bamboo flowering and livestock effects on understory composition and structure Structural characteristics of the understory community responded strongly to exclusion of herbivores.

- Interactive effects of introduced herbivores and post-flowering die-off of bamboos -

375

Fig. 2. (a) Chusquea culeou seedling density, (b) mean culm height.

Total cover increased steadily in fenced treatments but remained constant in unfenced plots (L × T: F3,36 = 20.59, P < 0.0001, Fig. 3a). This upward trend in total cover in fenced treatments was mostly due to the increase in herb cover (L × T: F3,36 = 48.88, P < 0.0001, Fig. 3b). By the year 2005, total understory cover was 76% and 36% in fenced and unfenced treatments, respectively (L: F1,12 = 25.80, P < 0.0001). In contrast, flowering condition per se did not significantly (P > 0.05) affect understory cover trends nor the final cover in 2005. However, over the 4-year period there was a slight tendency towards a steeper increase in total cover in flowered plots (ca. 36 to 90%) compared to non-flowered plots (ca. 37 to 63%; L × F: F1,12 = 3.06, P < 0.1, Fig. 3a), particularly when only herb cover was considered (L × F: F1,12 = 4.17, P < 0.07, Fig. 3b). Woody species cover by 2005 remained below 10% and was unrelated to herbivory or to bamboo flowering condition (P > 0.05, Fig. 3c). However, the richness of woody species increased more in the fenced than in unfenced plots (L × T: F3,36 = 5.92, P < 0.01). In terms of composition, MANOVA performed on cover of the 21 most abundant species as predictands and herbivory and bamboo flowering as predictors did not result in a significant model (P > 0.05). Univariate ANOVAs indicated that the only species that significantly responded to treatments was Alstroemeria aurea. By 2005, this species showed a 4-fold increase in cover in the fenced versus the unfenced bamboo-flowered treatment (L × F: F1, 12= 9.98, P < 0.01), but in non-flowered plots there was no significant difference between fenced and unfenced treatments.

Fig. 3. (a) Total non-arboreal understory cover (except Chusquea culeou); (b) understory herb cover; (c) woody species cover.

Discussion Effects on tree regeneration In both flowered and non-flowered bamboo patches, the rate of establishment of new seedlings of N. dombeyi was low. In this region, seedling establishment is abundant on sites that experience coarse-scale disturbance eliminating the understory and exposing bare mineral soil (e.g. severe fire, mass movements, logging and road construction). However, micro-sites in our study were covered in fine litter and partially shaded by living or dead bamboos. Although bamboo flowering appears to be sufficient to favor height growth of previously established N. dombeyi seedlings, the intensity of this disturbance does not appear to be sufficient to remove inhibitory influences associated with the understory, alive or dead, and trigger massive new seedling establishment. Even in the case of recruitment of previously established seedlings of N. dombeyi into taller height classes, the effects of livestock are contingent upon bamboo flowering, and appear to involve both direct effects of livestock and

376

Raffaele, E. et al.

indirect influences related to shifts in available forage. Browsing on tree seedlings was twice as great in the flowered as in the non-flowered unfenced plots (Fig. 2d). Tree seedlings were more heavily browsed in the absence of live bamboo, probably due to a dilution effect associated with greater alternative forage, and the difficulty of finding small tree seedlings in a dense mass of bamboo. This implies a positive indirect effect of live bamboo on N. dombeyi seedlings in the presence of introduced herbivores, contrasting with the expectation that bamboo inhibits tree regeneration due to shading. Still, direct suppression of tree seedlings by bamboo shading is also important, as indicated by the fact that in fenced plots N. dombeyi seedlings grew ca. 50% faster in flowered patches than in non-flowered patches. Veblen et al. (1989, 1992) also found that effects of introduced herbivores on tree regeneration are often mediated through reductions in more palatable forage species (e.g. Aristotelia chilensis). Depending on the intensity of herbivory, initial impacts may include increased densities of tree seedlings if they are released from severe understory competition, but continued herbivore pressure and a shift in diet towards tree species eventually inhibits tree regeneration. However, in contrast to previous studies of livestock impacts in Nothofagus forests during vegetative periods of bamboo, the current study is the first to examine herbivore impacts during a potential window of opportunity for tree regeneration. Livestock significantly reduced tree seedling growth in flowered bamboo patches and induced high mortality due to trampling or uprooting of seedlings irrespective of bamboo flowering. Overall, livestock browsing resulted in more effective inhibition of tree regeneration in flowered compared to non-flowered bamboo patches. In contrast with other studies from Asia (Nakashizuka & Numata 1982; Taylor 1992; Abe et al. 2001), bamboo flowering and die-off did not induce a pulse of new establishment of trees. Although seed production was not monitored at the site, it is unlikely that seed limitation explains the scarcity of seedlings. This is because N. dombeyi has been documented to produce large quantities (i.e. millions per ha) of seed at intervals of 2 to 3 years (Burschel et al. 1976; Donoso 1993; Veblen et al. 1996), and we observed some newly established seedlings in and near the study plots. Although N. dombeyi would be expected to establish and grow in high light environments, such as patches of flowered bamboo beneath canopy gaps, germination and/or early seedling survival may have been impeded by deep (5-10 cm) persistent litter layers produced by the bamboo withering. Deep litter is an unsuitable substrate for the small-seeded N. dombeyi, which generally germinates and establishes in greater abundances on mineral soils or on coarse woody debris than on forest floors of fine litter (Veblen et al. 1996).

These results are similar to the effects of massive flowering and die-off of other Chusquea species on regeneration of Nothofagus spp. in other forest types. Following massive flowering of C. quila in 1992-1993 in a low-elevation forest in south-central Chile, N. obliqua, which is small-seeded and shade-intolerant, competed poorly with new establishment of C. quila and experienced high rates of seedling mortality (González et al. 2002). The die-off of C. quila favored the growth of previously established seedlings and suckers of shadetolerant tree species, but under the low light levels of that forest type, advance regeneration of N. obliqua was absent. In the absence of a mast seeding year for N. obliqua soon after the bamboo flowering event, it is unlikely that this tree species will regenerate following the die-off of C. quila. Similarly, following a massive flowering of C. montana in 2001 in a subalpine forest of N. pumilio and N. dombeyi at ca. 40° S, Nothofagus seedlings did not establish in greater abundance at flowered than non-flowered sites (Holz & Veblen 2006). Instead, height growth and survival rates of previously established Nothofagus juveniles were enhanced in the flowered C. montana stands. These and the current study indicate that, following bamboo die-off, new establishment of shadeintolerant Nothofagus spp. is scarce and the dominant mode of tree regeneration consists of accelerated growth of juveniles established prior to the die-off event. Effects on understory Chusquea culeou seedlings established abundantly after the 2000 flowering event, reaching a peak of ca. 90 seedlings/m2 in 2003 and subsiding to ca. 65/m2 in 2005. Given the low (20%) overall Chusquea cover due to decades of browsing by livestock, we believe these seedling densities are much less than what would occur in the absence of livestock. New establishment of C. culeou occurred, indicating a capacity for germination from seed stored in the litter or soil. Thus, in addition to the widely documented semelparous reproductive cycle of most Chusquea spp., our study documents a viable seed bank that yielded new seedlings annually over the 4-year period of observation from seed produced during the flowering event in 2000. This multi-year seed bank strategy may enhance success of new seedling establishment in this region of highly variable year-to-year climate as well as in response to fluctuating populations of herbivorous small rodents. Despite chronic browsing and trampling before and during our four year study, C. culeou re-establishment does not seem to be heavily impacted by cattle. Despite an indication of reduction of bamboo seedling densities due to livestock effects, the densities attained of several 100 000s of seedlings/ha are sufficient for full recovery

- Interactive effects of introduced herbivores and post-flowering die-off of bamboos of the C. culeou population. Five years after flowering (2005) bamboo seedlings were 25.9 ± 3 cm tall, overtopping most understory shrub and herb species as well as tree seedlings. Changes observed in abundances and sizes of understory plants after fencing indicate that livestock strongly influence the herb component of the understory. Although these findings are consistent with studies conducted during the vegetative phase of C. culeou (Veblen et al. 1992), our results show that bamboo die-off intensifies livestock impacts for at least one species. Fencing in bamboo-flowered patches favored higher herb cover, particularly of pre-existing species such as the rhizomatous Alstroemeria aurea, compared to fenced plots in non-flowered patches. However, even under high light levels of flowered plots, herbivores also strongly limit herb dominance, either directly through consumption or by trampling. Prior to the introduction of large herbivores, periodic flowering of bamboo may have provided heliophyllous herbs such as A. aurea temporary windows of opportunity to gain dominance until the new bamboo thicket again reduced resources. Greater increases in total plant cover, herb cover, and richness of shrub species in fenced and flowered bamboo patches compared to unfenced flowered patches over the four years of our study indicate the potential of livestock for altering understory vegetation following the withering of the bamboo. Although determination of the full impact of livestock on abundance and richness of understory species would require longer term studies, under current levels of livestock, future reductions are likely in abundance and richness of native understory species. This expectation is based on the marked impacts observed during a single bamboo die-off event and the likelihood that only two or three similar bamboo flowerings have occurred in the region since livestock populations became abundant in the early 1900s. Long-term consequences of herbivores and bamboo flowering on forest dynamics Our results show that both cattle and die-off of flowered bamboos are major influences on the regeneration dynamics and understory vegetation of the N. dombeyi forest studied. The results of the current study are applicable to this forest type where N. dombeyi is the sole dominant and C. culeou is one of the understory dominants. Different responses to bamboo die-off would be expected further west where more shade-tolerant tree species are present, and further east, where conditions are drier and C. culeou gradually declines in importance. The results of our study apply to post-fire forests that have reached a stage of stand development in which treefalls provide gaps in which the shade-intolerant N. dombeyi

377

can regenerate. In contrast, in young post-fire forests with closed canopies, the regeneration of N. dombeyi is nil even in the absence of cattle and bamboo (Veblen et al. 1996). Due to widespread burning in northern Patagonia in the late 1800s-early 1900s and abundance of cattle in the forests since at least the early 1900s (Veblen et al. 2003), the stand conditions at L. Falkner are the most common stand situations encountered throughout the N. dombeyi forest type. In such stands, cattle have not only maintained a more open and shorter bamboo stratum but have also prevented N. dombeyi juveniles from attaining greater heights and densities beneath treefall gaps. Thus, in interpreting the results of the present study we must consider these legacies of a long history of cattle impacts. For example, the bamboo treatments (flowered versus non-flowered) are representative of patterns expected for stands that are suddenly released from chronic cattle impacts but might be somewhat different than patterns expected for more pristine forests. Specifically, in our study the contrast in light and potentially other resources such as soil moisture and nutrients between flowered and non-flowered bamboo patches was substantially less than that expected in a livestock-free forest with taller and more extensive bamboo. Thus, under pristine conditions the flowering and die-off of bamboo may engender a more intense response of the understory vegetation. Likewise, the greater abundances and sizes of juveniles of N. dombeyi expected for a livestock-free forest would probably enhance the chances of these species growing into the main canopy despite competition from newly established bamboo seedlings. Reduction of the size and abundance of N. dombeyi juveniles by previous herbivory diminishes the chances of successful tree regeneration after bamboo die-off even if cattle are excluded. Overall, our results imply that under current livestock loads in N. dombeyi forests, even following C. culeou flowering and die-off, tree regeneration beneath treefall gaps will continue to be slow and perhaps insufficient to maintain dominance by N. dombeyi. This interpretation is based on the slow rate of N. dombeyi seedling establishment and its high rate of seedling mortality, the reduction in size and abundance of advance tree regeneration by browsing during the vegetative phase of the bamboo, and especially the inhibition of tree seedling height growth during bamboo die-offs. Livestock inhibition of the height growth of N. dombeyi seriously reduces the chances of tree seedlings escaping the inhibitory influences of the rapidly growing bamboos.

378

Raffaele, E. et al.

Conclusions Massive bamboo flowering at intervals of ca. 60 years, potentially are important in creating long-lasting ecological patterns. Long-lasting legacies of rare, large-scale physical disturbances such as severe fire or windstorms are well recognized, but few studies have documented the importance of changes in forest ecosystem structures related to the life cycle of an understory plant species. Our study documents the importance of indirect interactions through changes in the availability of forage (e.g. bamboo death). Thus, prediction of herbivore impacts on the regeneration of particular tree species based on consideration of palatability alone may be misleading. Instead, assessments of herbivore impacts need to allow for indirect interactions and consider community-level approaches. Although rare bamboo flowering events formerly resulted in successful recruitment of previously established seedlings of the dominant forest species into the main canopy by allowing height growth releases during windows of reduced competition from bamboo, large introduced herbivores have dramatically impeded such regeneration. Thus, the interactive effects of introduced species with natural events such as bamboo flowering could have significant effects on ecosystem structure and composition. Acknowledgements. We thank J. Paritsis, C. Quinteros, M. Nuñez, R. Vidal Russel, G. Amico, N. Tercero, and C. Planas for field assistance, and the Administración de Parques Nacionales for permission to conduct the research. This study was supported by Agencia Nacional de Promoción Científica y Técnica (Grant PICT 01-09856), the National Geographic Society (Grant 7155-01), and the National Science Foundation of the USA (Award 0117366). ER, TK are researchers funded by Consejo Nacional de Investigaciones Científicas y Tecnológicas (CONICET, Argentina). References Abe, M., Miguchi, H. & Nakashizuha, T. 2001. An interactive effect of simultaneous death of dwarf bamboo canopy gap, and predatory rodents on beech regeneration. Oecologia 127: 281-286. Anderson, M.C. 1964. Studies of the woodland light climate I: The photographic computation of light conditions. J. Ecol. 52: 27-41. Burschel, P., Gallegos, C., Martínez, O. & Moll, W. 1976. Composición y dinámica regenerativa de un bosque virgen mixto de raulí y coihue. Bosque 1: 55-74 Correa, M.N. 1969-1984. Flora Patagónica. 4 Vols. Inst. Nacional de Tecnología Agropecuaria, Buenos Aires, AR. Donoso, C. 1993. Bosques templados de Chile y Argentina. Variación, estructura y dinámica. Editorial Universitaria S.A., Santiago, CL. González, M.E., Veblen, T.T., Donoso. C. & Valeria, L. 2002.

Tree regeneration responses in a lowland Nothofagus dominated forest after bamboo dieback in South Central Chile. Plant Ecol. 161: 59-73. Gurevitch, J. & Chester, T.S. 1986. Analysis of repeated measures experiments. Ecology 67: 251-255. Holz, C.A. & Veblen, T.T. 2006. Tree regeneration responses to Chusquea montana bamboo dieback in a subalpine Nothofagus forest in the southern Andes. J. Veg. Sci. 17: 19-28. Jaksic, F.M & Lima, M. 2003. Myths and facts on ratadas: Bamboo blooms, rainfall peaks and rodent outbreaks in South America. Austr. J. Ecol. 28: 237-251. Janzen, D.H. 1976. Why bamboos wait so long to flower? Annu. Rev. Ecol. Syst. 7: 347-391. Kitzberger, T. & Veblen, T.T. 2003. Influences of climate on fire in northern Patagonia, Argentina. In: Veblen, T.T., Baker, W.L., Montenegro, G. & Swetnam, T.W. (eds.) Fire regimes and climatic change in temperate ecosystems of the western Americas, pp. 290-315. Springer-Verlag, New York, NY, US. Nakashizuka, T. & Numata, M. 1982. Regeneration process of climax beech forest I. Structure of a beech forest with undergrowth of Sasa. Jpn. J. Ecol. 32: 57-67. Pearson, A.K., Pearson, O.P. & Gomez, I.A. 1994. Biology of the bamboo Chusquea culeou (Paceae: Bambusoideae) in southern Argentina. Vegetatio 111: 93-126. Relva, M.A. & Caldiz, M. 1988. Composición estacional de la dieta de ciervos exóticos en la Isla Victoria, Parque Nacional Nahuel Huapi. Gayana Zool. 62: 101-108. Taylor, A.H. 1992. Tree regeneration after bamboo die-off in Abies-Betula forest, Wolong Natural Reserve, China. J. Veg. Sci 3: 253-260. Vázquez, D.P. 2002. Multiple effects of introduced mammalian herbivores in a temperate forest. Biol. Invasions 4: 175-191. Veblen, T.T. 1989. Nothofagus regeneration in treefall gaps in northern Patagonia. Can. J. For. Res. 19: 365-371. Veblen, T.T. 1982. Growth patterns of Chusquea bamboos in the understory of Chilean Nothofagus forests and their influences in forest dynamics. Bull. Torrey Bot. Club 109: 474-487. Veblen, T.T., Donoso, C., Kitzberger, T. & Rebertus, A.J. 1996. Ecology of southern Chilean and Argentinean Nothofagus forests. In: Veblen, T.T., Hill, R.S. & Read, J. (eds.) Ecology and biogeography of Nothofagus forests, pp. 293-353. Yale University Press, New Haven, CT, US. Veblen, T.T., Kitzberger, T., Raffaele, E. & Lorenz, D.C. 2003. Fire history and vegetation change in northern Patagonia, Argentina. In: Veblen, T.T., Baker, W.L., Montenegro, G. & Swetnam, T.W. (eds.) Fire regimes and climatic change in temperate ecosystems of the western Americas, pp. 259-289. Springer-Verlag, New York, NY, US. Veblen, T.T, Mermoz, M., Martín, C. & Ramilo, E. 1989. Effects of exotic deer on forest regeneration and composition in northern Patagonia. J. Appl. Ecol. 26: 711-724. Veblen, T.T., Mermoz, M., Martín, C. & Kitzberger, T. 1992. Ecological impacts of introduced animals in Nahuel Huapi National Park. Conserv. Biol. 6: 71-83. Received 18 April 2006; Accepted 13 December 2006; Co-ordinating Editor: A. Symstad.