BOSQUE 32(1): 64-76, 2011

ARTÍCULOS

Effect of a tourist-ignited wildfire on Nothofagus pumilio forests at Torres del Paine biosphere reserve, Chile (Southern Patagonia)

Efecto de un incendio forestal causado por un turista sobre bosques de Nothofagus pumilio en la reserva de la biosfera Torres del Paine, Chile (Patagonia Austral)

Osvaldo J Vidal ^a*, Albert Reif ^a

* Corresponding author: ^a Albert Ludwigs University Freiburg, Faculty of Forest and Environmental Sciences, Silviculture Institute, Tennenbacherstrasse 4, D 79106 Freiburg, Germany, tel.: (49) (0761) 2033675, fax: (49) (0761) 2033781, osvaldo.vidal@waldbau.uni-freiburg.de.

SUMMARY

This study describes the impact of a wildfire caused by a tourist at Torres del Paine Biosphere Reserve in 2005, which destroyed approximately 2,000 ha of forests. Changes on biodiversity attributes of Nothofagus pumilio stands including structure, floristic composition and regeneration, are shown. We compared three plot groups having similar site and structure: Cordón Masle burned plots (CMBP); Cordón Masle unburned plots (CMUP), adjacent to the first ones; and Cerro Paine unburned plots (CPUP), outside the fire perimeter. Data was analyzed using univariate (Mann-Whitney) and multivariate (MRPP and ISA) non-parametric analysis. In CMBP, the fire reduced in average 91.7 % of the tree density and 70.9 % of the basal area. In the unburned plots the average of seedling was 23,625 and 22,750 ha^-1 in CPUP and CMUP respectively, whereas in burned ones it was only 1,125 ha^-1 in CMBP. Canopy layer collapsed and species richness and cover on ground increased significantly in burned plots. The average of exotics increased from 1.5 and 0.6 in CPUP and CMUP respectively to 11.9 in CMBP. Tree mortality, lower regeneration, biological invasion and broken canopy-ground balance in burned stands, are altering the successional pathways of forests, making difficult the forest regeneration and restoration efforts. Based on this information, we propose managers to encourage efforts in fire prevention and to incorporate a database on biodiversity indicators for management plans, in order to know more about the ecological integrity of ecosystems and the impacts provoked by recreation.

Key words: biodiversity indicators, biological invasions, fire ecology, forest structure, human disturbances, recreation.

RESUMEN

Se describe el impacto de un incendio causado por un turista en la reserva de la biosfera Torres del Paine en el 2005 que destruyó aproximadamente 2.000 ha de bosques. Se muestran los cambios ocurridos en atributos de bosques de Nothofagus pumilio incluyendo estructura, regeneración y composición florística. Se comparan tres grupos de parcelas con sitio y estructura semejantes: parcelas incendiadas (CMBP) y parcelas adyacentes no incendiadas (CMUP) del Cordón Masle, y parcelas no incendiadas de Cerro Paine (CPUP), fuera del perímetro del incendio. Se utilizó análisis no paramétricos univariados (Mann-Whitney) y multivariados (MRPP e ISA). En los sitios incendiados, el fuego redujo en promedio el 91,7 % de la densidad arbórea y el 70,9 % del área basal. El promedio de plántulas fue de 23.625 y 22.750 ha^-1 en CPUP y CMUP, respectivamente, mientras que en CMBP se redujo a 1.125 ha^-1. El dosel colapsó y la cobertura y riqueza de especies de sotobosque incrementó significativamente en las parcelas incendiadas. El promedio de exóticas incrementó de 1,5 y 0,7 especies en CPUP y CMUP, respectivamente, hasta 11,9 especies en CMBP. La mortalidad de árboles, escasa regeneración, invasión biológica y el colapso del dosel, están alterando las vías sucesionales de los bosques reduciendo su capacidad de recuperación, dificultando los esfuerzos de restauración. Se propone a los administradores enfatizar esfuerzos en prevención de incendios y desarrollar información sobre indicadores de biodiversidad para los planes de manejo, para así conocer más sobre la integridad ecológica de los ecosistemas y el impacto provocado por la recreación.

Palabras clave: indicadores de biodiversidad, invasiones biológicas, ecología del fuego, perturbaciones humanas, recreación.

INTRODUCTION

Wildfires affecting Andean Nothofagus spp. forests in southern Patagonia are recognized as catastrophic disturbances, which have strong influences on diversity attributes by changing abruptly the states of dynamics in these ecosystems (Veblen et al. 2008). Beyond their anthropic or natural origin, wildfires produce changes that affect critical components of ecosystem processes like soil properties, vegetation layering patterns and under-story species composition. Fire consequences imply among others, reduction in levels of organic matter on soils (Certini 2005), increment of incoming solar radiation on under-story (Kitzberger et al. 2005), and increased competition because of the income of exotic species in the original ensemble of native ones (Freeman et al. 2007, Donato et al. 2009). Although Andean Nothofagus spp. forest ecosystems seem to be able to recover after wildfires in absence of ulterior disturbances (Veblen et al. 1996), the relative increment in frequency and severity of wildfires by Euro-Chilean settlers (González et al. 2005) appears to exceed the critical thresholds of resilience in many of these ecosystems, especially in forest dominated by species lacking regenerative strategies after wildfires (Veblen et al. 2008). Thus, a rapid conversion from forests to more open vegetational landscapes like shrub-lands and grassland areas has been carried out during the last century through land use practices, livestock production and, more recently, through ex-urban population and recreation (Martinic 1985, Veblen and Markgraf 1988, Veblen et al. 2008).

In a context of biological conservation, to know how wildfires affect structure, composition and function of biodiversity at different levels of organization (Noss 1990), and at which levels these disruptions are capable of exceeding resilience thresholds, are becoming key questions for ecological management and restoration purposes (Noss 1990). Such questions are particularly interesting to be addressed in wilderness areas, which are devoted to protecting natural ecosystems and to serving as reference systems of ecological integrity from human impact (Cole and Landres 1996). In Chile wilderness areas are increasingly subjected to anthropogenic pressure arising from recreation and settlement in surrounding areas (Pauchard and Villarroel 2002), thus the generation of detailed information on biodiversity patterns is crucial because this can offer indicators as reference of pristinity or can give light on the ecological significance of potential or real impacts of disruptions (Noss 1990, Cole and Landres 1996). Although generalizations from literature could be useful by providing some guidelines for managers, local studies on this oriented focus seem to be more pertinent because of their assertiveness. This is particularly true for developing ecosystems that present wide range of distribution like the Nothofagus pumilio (Poepp. et Endl.) Krasser forests, which are distributed along the Andean for approximately 2,000 km and present a wide structure and compositional range (Hindelbrand-Vogel et al. 1990). On 17 February 2005, a huge wildfire was ignited at Torres del Paine Biosphere Reserve (Southern Chilean Patagonia), when a careless tourist accidentally flipped his stove on ground vegetation. The fire of approximately four weeks of duration, began on a grassland area but spread rapidly towards other communities, destroying valuable habitats including steppes, shrub-lands and forests along 17,000 ha inside the reserve and adjacent private lands, including approx. 2,000 ha of N. pumilio forests (Domínguez et al. 2006, Navarro-Cerrillo et al. 2008). Although the effects of this wildfire event have been evaluated from a landscape perspective using remote sensor data to estimate fire perimeter and affected surfaces (Navarro-Cerrillo et al. 2008), no studies have been focused on showing the ecological effects of this disturbance from a more biodiversity detailed approach. Given the importance of this Biosphere Reserve as a national symbol of Chilean conservation areas, the development of more detailed information on the ecological impacts of this wildfire event appears to be an issue of relative magnitude which could help for conservation and restoration strategies in management plans. Thus, the objectives of the present paper are: 1) to develop biodiversity indicators as reference attributes of N. pumilio forests in this Biosphere Reserve and 2) to show the effect of fire on three sensitive ecosystem components belonging to these forests: stand structure, regeneration and floristic composition.

METHODS

Description of the study area. Fieldwork was conducted at Torres del Paine Biosphere Reserve, a protected area located in Última Esperanza province in southern Chilean Patagonia. This 242,000 ha area is dissected by a complex physiography characterized by the Southern Patagonian Ice Fields and the Cordillera Paine, an impressive mountain range that reaches abruptly over 3,000 m. Such contrasting relief determines a variety of climates and an abrupt moisture gradient in west-east direction fluctuating from 1,000 to 300 mm in annual precipitation (Pisano 1974). The mean temperature reaches 10.8 °C in summer (January) while in winter season (July) reaches 2 °C. Bio-climatologically the entire area was classified as belonging to the humid-supratemperate belt (Amigo and Ramírez 1998). The soils are from quaternary origin after the glacier retreated approximately 10,000 years ago (Armesto et al. 1992), dominating the Andisol soil types with deposits of volcanic origin (Navarro-Cerrillo et al. 2008). The terrestrial vegetation is diverse, and ranges from arid-steppe communities to mixed Nothofagus spp. forests in the moistest places (Pisano 1974). Pure stand forests dominated by N. pumilio are located on mountain slopes and valleys, on intermediate elevation levels from 300 to 900 m altitude and with also intermediate precipitation regimes from 400 to 900 mm of annual precipitation (Pisano 1974, Armesto et al. 1992). The area was declared a UNESCO Biosphere Reserve in 1978 because of the beauty of its landscapes and its ecosystem representativeness (Pisano 1974, CONAF 2007).

Plot sampling selection. Two locations were selected as study areas to compare the effects of fire on forest components: Cordón Masle mountain range located in the northwest limit of the natural reserve, and Cerro Paine mountain range located approximately 9 km south-east from the first one. In Cordón Masle the fire perimeter affected extensive forested areas of Nothofagus spp. during the event of 2005 (Navarro-Cerrillo et al. 2008). This area (Cordón Masle) is a middle-elevation mountain range that reaches up to 900 m a.s.l., with mixed stands of N. antarctica (G. Forster) Oersted and N. pumilio in lower areas close to watersheds and pure stands of N. pumilio in elevation from approximately 250 m up to 800 m. Above this elevation, forests are progressively replaced by Krummholz (Pisano 1974). Forested areas here developed from gentle to steep slopes with south and southeast exposure. Cerro Paine was not located within the fire perimeter of the 2005 event (Navarro-Cerrillo et al. 2008). This mountain range reaches up to 1,200 m, containing pure forest stands from 300 to 800 m. Similarly to Cordón Masle forested areas, pure N. pumilio forest stands here develop both on gentle and steep slopes with south and south-east exposure.

In February and March 2009 a total of 24 plots in pure N. pumilio stands were installed on these two localities according to a stratified sampling, in order to form three comparison groups: 1) Cordón Masle burned plots (CMBP), consisting of a group of eight plots burned on slopes of south and south-west exposure between 381 and 638 m in elevation; 2) Cordón Masle unburned plots (CMUP), consisting of a group of eight unaffected plots immediately adjacent to burned ones in Cordón Masle, located on slopes of south and south-west exposure, from 430 to 612 m in elevation. 3) Cerro Paine unburned plots (CPUP), consisting of a group of eight unburned plots outside the fire perimeter in Cerro Paine mountain range and from 356 to 674 m (figure 1).

The criteria for plot selection were as follows: 1) burned stands were selected having a minimum size to establish a 500 m² stand plot; 2) after establishing the burned plot, an immediately unburned contiguous forest stand was selected. Here it is assumed that stand structure (e.g. canopy height or trunk diameter) and site conditions (e.g. elevation, slope classes, aspect and soil texture) are similar because of proximity between both stands; 3) a third group of plots in Cerro Paine was installed; 4) un-burned plots showed low or no evidence of disturbances by large herbivores (domestic livestock and guanaco (Lama guanicoe)), based on counting of browsed branches inside the plots (Veblen et al. 1989) (appendix 1).

Stands plot structure recordings and regeneration counts. Forest structure was characterized by installing a rectangular 50 x 10 m plot. In each plot from each group, diameters at breast height (DBH) were recorded on all trees having > 5 cm diameter at 1.3 m high. Because almost greater size-classes trees in burned plots from Cordón Masle remained standing, it was also possible to record DBH of such dead trees. This situation allowed us to generate a direct estimation on damages to tree density and basal area, by partitioning the Cordón Masle burned plots (CMBP) into two sub-groups, namely: a) Cordón Masle burned plots a (CMBPa), which includes assessments on live and dead trees and b) Cordón Masle burned plots b (CMBPb), where only live trees are considered for such estimations. Regeneration was assessed on each plot by counting the number of seedlings having between 1 and 15 cm tall.

Vegetation sampling. In the center of each stand, one vegetation subplot of 10 x 10 m was delimited in order to record floristic composition. Percent coverage of each vascular plant was recorded by visual estimation. All species were attributed to a particular horizontal layer on recordings, namely: a) stratum 1 of species present in under-story layer (0 – 0.5 m in height); b) stratum 2 of species present in the low shrub layer (0.5 – 1 m in height); c) stratum 3 of species present in the high shrub layer (1 - 5 m in height) and d) tree layer (> 5 m in height). Species not determined in situ were collected for laboratory identification using taxonomical literature (Moore 1982, Matthei 1995, Rodríguez and Quezada 2003). Nomenclature follows to Henríquez et al. (1995) and Domínguez et al. (2006).

Database analyses. Data on stand structure and regeneration were used to elaborate a database for comparisons amongst the groups of plots. We estimated total tree density (including both live and dead standing trees in burned plots), live standing tree density, tree DBH, total basal area (including both live and dead standing trees in burned plots), live standing basal area and regeneration (saplings and seedlings). Data on floristic composition was used to estimate means and medians for richness and cover of species based on groups of common features among species. The following features were assigned for each species: phytogeographic origin (natives or non-native species) (Matthei 1995, Henríquez et al. 1995, Domínguez et al. 2006), general life forms (trees, tall shrubs, low shrubs, herbs, graminoids, epiphytes) (Donato et al. 2009), and Raunkiaer’s plant life forms (phanerophytes, chamaephytes, hemicryptophytes, therophytes, geophytes, epiphytes) (Mueller-Dombois and Ellenberg 1974). In addition, Shannon and Berger Parker indexes were computed for quantifying differences on diversity and taxonomical homogeneity among the groups. These indices were computed from the sum of all absolute covers of individual plant species in each group.

Statistical analyses. Stand structure (tree densities, basal area, dbh, and vegetation layers), regeneration (number of seeds and saplings), plant life forms (general and Raunkiaer’s system) and phytogeographic origin (native and exotic plants) were compared among groups using Mann-Whitney U-tests. Statistic significance was accepted at the 0.05 P level. Shannon and Berger-Parker indexes were compared by generating confidential intervals from bootstrap estimators (1,000 iterations). Univariate analyses were performed using the PAST software (Hammer et al. 2001).

"Multi-response permutation procedure" (MRPP) using Euclidean was performed to test the hypothesis of no difference in species composition among groups. In this test, an A statistics describes the degree of within-group homogeneity compared to that expected by chance (called effect size) (McCune and Grace 2002). When all entities within groups are identical, A reaches its maximum value (A = 1). When the heterogeneity within groups equals the level expected by chance, then A = 0, and when there is more heterogeneity within groups than the level expected by chance, then A < 0. In MRPP a P value is given for each test group comparison. Pre-defined groups of plots as described above were used to make multiple comparisons based on species composition data. Following the MRPP, we performed "indicator species analyses" (ISA) (Dufrêne and Legendre 1997) to verify fidelity of species to burned and unburned conditions. These species are responsible for differences between groups in the MRPP. ISA yields an indicator value and a statistical significance for this value from Monte-Carlo randomizations. PC-ORD software (MacCune and Grace 2002) was used for these multivariate analyses.

RESULTS

Responses on structural attributes. Differences among groups on tree density and basal area were statistically significant. Average densities (± SD) of living trees diminished from 545 ± 251 trees ha^-1 (CMBPa) to 45 ± 59 trees ha^-1 (CMBPb) (Mann-Whitney U-test: z = -3.31, P < 0.001), while the basal area of living trees was reduced from 51.9 ± 16.9 m² ha^-1 (CMBPa) to 15.1 ± 13.0 m² ha^-1 (CMBPb) (Mann-Whitney U-test: z = -3.10, P < 0.001). These changes were caused by high fire intensity, which destroyed 91.7 % of the initial tree densities and 70.9 % of the basal area of evaluated stands. Tree mean diameters, however, remained unchanged (figure 2).

Vegetation covers were changed after the fire (table 1). On average, ground vegetation in CMBP reached 96.3 ± 45.8 % of relative cover, whereas in CPUP and CMUP this value reached 52.1 ± 29.4 % and 59.1 ± 12.7 %, respectively (Mann-Whitney U-test, P < 0.001). In contrast, tree cover layer reached a reduced value in CMBP of 1.5 ± 2.3 %, though in CPUP and CMUP this value reached higher values of 37.3 ± 11.7 % and 43.0 ± 5.3 %, respectively (Mann-Whitney U-test, P < 0.001). Shrub layer also shows significant differences among groups. Thus, in CMBP shrub cover reached just 1.9 ± 2.6 %, while in CPUP and CMUP this value reaches 8.2 ± 9.4 % and 8.6 ± 6.4 %, respectively (Mann-Whitney U-test, P < 0.001) (table 1).

Differences on regeneration patterns were also statistically significant among groups (Mann-Whitney U-test, P < 0.001). On average 1,125 ± 1,123 seedlings ha¹ were found in CMBP, while in CPUP and CMUP 23,625 ± 24,908 and 22,750 ± 16,901 seedlings ha¹ were found, respectively (table 1).

Responses on species composition and diversity. Species composition varied significantly among burned and unburned groups for all species covers (MRPP: A = 0.253, T = -12.397, P < 0.001). Burned and unburned plot groups differed significantly (pairwise comparisons), but no differences were detected between the CPUP and CMUP (MRPP: A = 0.045, T = -1.204, P > 0.05). Indicator values computed by ISA showed that from a total of 77 species (appendix 2), 33 can be considered as belonging to burned or unburned conditions, with fidelity values fluctuating from 37.5 (relatively high) to 100 (absolute fidelity in any group). In unburned plot groups, seven species can be considered as belonging to this undisturbed condition (all of them natives) while 26 species are typical for burned condition (46 % of them native and 54 % exotic) (figure 3). The richness and cover of species showed also significant differences among groups (Mann-Whitney U-test, P < 0.001) (figure 4). The mean number of species (± SD) in CMBP was 28.0 ± 4.8, while in CPUP and CMUP it reached 17.9 ± 3.0 and 12.4 ± 3.7 respectively. By dividing this measure between native and exotic species, richness of native ones resulted significantly greater by comparing CPUP and CMUP 16.3 ± 2.7 v/s 11.8 ± 2.9 species) (Mann-Whitney U-test, z = -2.596, P < 0.001), while the comparison between CPUP and CMBP and also CMUP and CMBP remained not significant (Mann-Whitney U-test, P > 0.05). Exotic species reached significantly higher values in burned plots (Mann-Whitney U-test, P < 0.001). On average, exotic species richness reached values of 11.9 ± 4.2 in CMBP compared with the 1.5 ± 1.9 and 0.6 ± 1.1 obtained from CPUP and CMUP, respectively (Mann-Whitney U-test, P < 0.001). Considering the cover values, no significant changes among the groups were observed (Mann-Whitney U-test, P > 0.05). The absolute cover of species in CMBP reached values of 136 ± 46 %, while in unburned plots it reached 120 ± 18 % and 146 ± 18 % in CPUP and CMUP, respectively. Native species had significantly lower cover values in CMBP (73 ± 24 %) than in unburned ones (110 ± 16 % and 142 ± 20 % in CPUP and CMUP, respectively) (Mann-Whitney U-test, P < 0.01). Exotic species however, which were practically absent in CPUP (10 ± 13 %) and CMUP (4 ± 8 %), reached higher values in CMBP (62 ± 33 %) (P < 0.001) (figure 4).

Berger-Parker index (d) shows lower values for the CMBP (d = 0.1614, CI = 0.1421 – 0.1827) than those permuted for CPUP (d = 0.3641, CI = 0.333 – 0.3951) and CMUP (d = 0.4227, CI = 0.392 – 0.4552), indicating the reduced cover values of the dominant species (N. pumilio) in burned plots. Shannon index (H’) also showed greater values in CMBP (H’ = 3.212, CI = 3.105 – 3.254) than those in CPUP (H’ = 2.322, CI = 2.211 – 2.384) and CMUP (H’ = 1.859, CI = 1.763 – 1.918) showing the incorporation of new species to the disturbed places and relative homogeneity in cover values among them (table 1).

Responses of plant life forms. Both number and cover of phanerophyte and epiphyte species had reduced values in burned plots (figure 5). The damage produced by fire destroyed the components present in the canopy, allowing the incorporation of new understory species to the original ensemble. Average of phanerophyte species (± SD) in CMBP reached values of 1.25 ± 1.4 compared with the 2.8 ± 1.2 and 3.1 ± 0.6 species appearing in CPUB and CMUP, respectively. Cover values in CMBP reached 2.5 ± 3.4 %, while in CPUP and CMUP these values reached 46.6 ± 14.7 % and 48.7 ± 5.1 %, respectively. By contrast, the richness and cover in burned places of both hemicriptophytes and therophytes increased significantly. Hemicriptophyte species incremented in average from 1.2 ± 1.0 and 0.7 ± 1.0 in CPUP and CMUP to 7.1 ± 2.7 in CMBP. Cover incremented from 23.4 ± 17.6 % and 21.4 ± 15.3 % in CPUP and CMUP respectively to 49.7 ± 15.9 % in CMBP. Therophyte species average increased from 1.1 ± 1.0 and 0.2 ± 0.5 species (CPUP and CMUP) to 4.7 ± 2.0 species in burned plots from Cordón Masle. Covers also increased from 6.1 ± 15.1 % and 0.5 ± 1.2 % (CPUP and CMUP) to 11.1 ± 6.4 % in burned plots (figure 5).