Vegetation responses to past volcanic disturbances at the Araucaria araucana forest‐steppe ecotone in northern Patagonia

Abstract Volcanic eruptions play an important role in vegetation dynamics and its historical range of variability. However, large events are infrequent and eruptions with a significant imprint in today's vegetation occurred far in the past, limiting our understanding of ecological processes. Volcanoes in southern Andes have been active during the last 10 ka and support unique ecosystems such as the Araucaria–Nothofagus forest. Araucaria is an endangered species, with a fragmented distribution and well‐adapted to fire and volcanic disturbances. Yet, it was suggested that volcanism might have increased the fragmentation. Through the use of pollen and tephra analysis from a sedimentary record, this paleoecological study aims to provide an insight into the vegetation responses to past volcanic disturbances, to assess the role of volcanic disturbance on the vegetation dynamics and to determine if the current fragmentation has been caused by volcanism. Results show that during the last 9 kyr, 39 tephra falls buried the vegetation around Lake Relem, more frequently between 4 and 2 ka. The pollen percentage indicates that the vegetation changed after small tephra fall but seldom caused significant changes. However, the large eruption of Sollipulli volcano (~3 ka) changed the environmental conditions affecting severely the vegetation. Ephedra dominated the early successional stage, perhaps facilitating Nothofagus recovering after ~500 years. Slight increase of Araucaria and Nothofagus obliqua‐type pollen percentages suggests that forest resisted without permanent changes and recovered relatively fast after the large eruption, perhaps because of sparse biological legacies distributed in the landscape. In the study area, the relative stability of Araucaria pollen after several tephra fall suggests no change in its past distribution at the current forest‐steppe ecotone, thus not affecting its current conservation status. Perhaps, random factors, the colonization patterns of the high elevations in the Andes after deglaciation and topography might play a more important role than previously thought.


| INTRODUC TI ON
Volcanic eruptions are important disturbance agents on Earth, and the impacts can trigger sudden and large environmental changes, particularly on the vegetation (Crisafulli & Dale, 2018). Large volcanic eruptions are infrequent, episodic, and stochastic events in the history of an ecosystem (Turner et al., 1998). Volcanism creates complex patterns and dynamic processes on the vegetation depending on the disturbance mechanism and the magnitude (e.g., Crisafulli & Dale, 2018;Foster et al., 1998). For example, after the eruption of Mount St. Helens in North America, the vegetation recovered rapidly after being buried by ashfall (Zobel & Antos, 2018), while literature suggests that the type and arrange of biological legacies in disturbed areas drove regeneration patterns and the rate of vegetation development after disturbance (del Moral & Grishin, 1999).
Understanding the lasting effects of volcanism on vegetation responses like the resistance, recovery or the development of diversity patterns may help in designing nature management (Dale et al., 2005;Franklin et al., 2002) especially in highly diverse biogeographical areas influenced by volcanism and affected by current climate change. However, the study of the vegetation dynamics is challenging since active volcanoes are located in remote areas, the long-term monitoring requires many resources, and the responses last too long to be monitored in human life span or had occurred far in the past (Swanson & Crisafulli, 2018) under different climate condition. For this end, well-designed paleoecological studies, through the study of tephra layers and pollen in sedimentary records, can help to understand the consequences of past volcanic disturbances on vegetation responses or can be used as an analogue for future events (e.g., Payne & Egan, 2017).
Paleoecological studies in western Europe, North-America, and New Zealand have attempted to illustrate past impacts of eruptions and aimed to contribute to disturbance ecology; however, results are not always consistent or show clear patterns. For example, palynological studies of large volcanic eruptions in Iceland and their effects on past climate showed contradictory or weak evidences that volcanoes disturbed the vegetation and past climate conditions in western Europe (Payne et al., 2013). Also in the Mt. Mazana, western North America, studies were conducted to infer the impacts on local and regional vegetation (e.g., Egan et al., 2016). The authors concluded eruptions did not trigger significant changes in terrestrial pollen taxa in a distant area, although noticeable increase in pollen dominance from Cupressaceae and Tsuga heterophylla following the eruption, while the aquatic taxa changed because of enrichment of nutrients.
However, after the Taupo eruption (~1850 years BP), New Zealand, pollen evidences show that this eruption destroyed the surrounding forest, and up to 170 km east of the vent the vegetation suffered a variable degree of disturbance (Wilmshurst & McGlone, 1996). In many areas, the forest could not recover to its original conditions, and sites far from the crater covered by thin tephra layer were strongly impacted.
Active volcanoes during the Holocene are widely dispersed on the Earth, but one of the most active areas is located in the south-eastern part of the Pacific (Stern, 2004). The subduction of the Nazca plate underneath the South American plate has triggered several large eruptions in the last 10 kyr in the Andes region (Fontijn et al., 2014). The southernmost section of the Andes supports unique forest ecosystems, characterized by a high endemism, which also has been influenced by these volcanoes eruptions affecting structure and functions in this Andean forest and surrounding vegetation (e.g., Veblen et al., 2016). As a result of the eruptions, the tree-line location can be depressed (Daniels & Veblen, 2004;Veblen et al., 1977) or sustain uneven age forest in the landscape (Kitzberger, 2012). Volcanism was thought to have been responsible for keeping pioneer species such as Nothofagus species and Araucaria araucana dominant (Burns, 1991;Veblen & Ashton, 1978). The evidences showed that after the eruption of the Puyehue-Cordon del Caulle Volcanic Complex (PCC) in 2011 Nothofagus pumilio was the principal species resprouting in zones buried by ~50 cm tephra, and the occurrence of several cohorts would correspond to past eruptions (Montiel et al., 2017). Eastward from PCC in 2011, the steppe vegetation was buried by <5 cm tephra and rhizomatous geophytes species such as Poa spp. and Rumex acetosella increased while therophytes disappeared (Ghermandi et al., 2015). The 2008-eruption of Chaitén volcano was also studied. Swanson et al. (2013)

T A X O N O M Y C L A S S I F I C A T I O N
Biogeography, Community ecology, Conservation ecology regeneration dynamics (e.g., Bekessy et al., 2002;Veblen et al., 2016).
Yet the hypothesis is contradictory with the recognized morphological adaptation of this species (e.g., thick bark and flexible branches) to resist volcanic disturbance (Burns, 1991) that might have helped to persist in an angiosperm-dominated ecosystems (Kershaw & Wagstaff, 2001). Despite Araucaria's socio-ecological importance and its current endangered conservation status, few long-term vegetation reconstructions have been conducted so far in the Araucaria forest-steppe ecotone. To contribute to the knowledge of vegetation dynamics and the volcanic disturbance regime, this work aims (1) to reconstruct the volcanic disturbance history in the Araucaria forest-steppe ecotone, (2) to assess the vegetation responses to past volcanic disturbance, and (3) to evaluate if Araucaria distribution was influenced by volcanism at the current forest-steppe ecotone affecting its conservation status. To this end, I reanalyzed a published pollen record (Moreno- Gonzalez et al., 2021) to have an insight into the volcanic influence on the Araucaria forest dynamics. Due to the distance of the volcanic source of the forest-steppe ecotone eastward of the Andes (~20-50 km), it is affected mostly by tephra fall that buried the vegetation. Tephra fall therefore is the main volcanic disturbance type around the study area. Following this assumption, I predict that the magnitude of vegetation responses should be related to the tephra thickness and that vegetation would be more resistant and/or recover faster to thin tephra fall than to thick tephra fall. In consequence, forest fragmentation should increase after thick tephra fall.

| ME THODS
The study area is located in the current Araucaria forest-steppe ecotone around 39° S, in northern Patagonia ( Figure 1). Vegetation history was reconstructed from a sediment core obtained from Lake Relem (38°58′39″S; 71°4′51″W; 1268 m a.s.l.). Lake Relem is a shallow and small lake (~2.5 m depth, ~1 ha, respectively), without river inflow or outflow. The climate in the region is temperate with oceanic influence. The Pacific air masses arriving with westerly winds create a sharp rain-shadow effect discharging most of the precipitation on the western section of the Andes (Mundo et al., 2013).
Climatic conditions when interacting with local topography and natural disturbances generate complex vegetation patterns and plants associations (Kitzberger, 2012;Roig, 1998). Araucaria forest occurs mainly above 1000 m elevation, normally reaching the tree-line at 1700 m a.s.l. (Gonzalez et al., 2006) showing a strong fragmentation ( Figure 1). Further details on climate and vegetation description, as well as a full description of the sediment analysis, pollen analysis, and the chronology can be found in Moreno-Gonzalez (2020) and Moreno- Gonzalez et al. (2021). In brief, the sediment core was dated with 7 AMS Radiocarbon dates and then the chronology was made with smooth spline (0.1 spar) at 95% confidence. At least 400 pollen grains were counted in each of the 176 pollen samples. The pollen samples were obtained every 2-4 cm, most of them in the following first 1-4 cm above the tephra layers. Time resolution per sample varied between ~−20 year/cm above the So-A tephra layer and ~40 year/cm below the So-A tephra layer.

| Volcanic setting and eruptive history
The study area lies in the Southern Volcanic Zone of the Andes (SVZ).
The SVZ is a result of the subduction of the Nazca plate beneath the continental South American plate and is extended between 33°S and 46°S (Gilbert et al., 2014). In a radius of 100 km around Lake Relem, there are seven volcanoes that have been active during the Holocene ( Figure 1). Fontijn et al. (2014) list all eruptions recorded so far to the SVZ, which are summarized in Table 1. The Lonquimay volcano has had more regular eruptions of considerable magnitude (Volcanic Eruption Index ≥ 3) while other volcanoes have erupted every 10 years (e.g., Villarrica volcano) or have erupted at least once during the last 10 k years (Fontijn et al., 2014). The Sollipulli-Alpehue eruption (hereafter the So-A eruption) was one of the largest in the recent past in northern Patagonia, dated back to 2951 cal year BP (Naranjo et al., 1993). Lake Relem is located about 40 km from the Sollipulli volcano within the tephra isopach of 2 m of the So-A eruption (Fontijn et al., 2016).

| Data analysis
For this study, I used pollen percentage and pollen accumulation rate (PAR) data from terrestrial taxa following standard method (Bennett & Willis, 2001). Calculation and methodology were further described in Moreno- . A stratigraphically constrained cluster analysis (CONISS) of the pollen samples (Grimm, 1987) was conducted based on Euclidean distances and calculated the amount of significant cluster zones by broken stick method (Bennett, 1996).
To assess pollen diversity changes, individual rarefaction analysis is a powerful tool (Birks & Line, 1992). Using rarefaction analysis, we estimated the palynological richness at the minimum terrestrial pollen count of 400 pollen grains (E(T 400 )), the pollen diversity estimated at E(T 10 ), and pollen evenness was calculated as the ratio E(T 10 )/E(T 400 ) (Matthias et al., 2015). Compositional trend of the terrestrial taxa was explored through principal component analysis. The percentage was square root transformed and centralized. Furthermore, we fitted a principal curve (PC) to the compositional data. The starting point was based on the age of the samples, and the curve was fitted through a smooth-spline method with complexity of 5. To estimate the rate-of-change (RoC), we interpolated the pollen samples at a regular time interval of 50 years with smooth spline and then with the Euclidean distances as a dissimilarity coefficient (Bennett & Humphry, 1995).
To reconstruct the volcanic disturbance regime, I made use of every tephra layer >0.5 cm thick, which was considered arbitrarily as a clearly deposited layer and not mixed with the sediment. Here I considered each of these tephra layers as a single and independent disturbance event-since tephra fall is limited to TA B L E 1 Location of potential source volcanoes close to Lake Relem, their frequency during the Holocene and amount of eruptions of considerable magnitude (VEI: Volcanic Eruption Index; further details of dates and uncertainty, chemical composition, references, for each eruption, are found in Fontijn et al., 2014).

Volcano name
Distance and direction from Lake Relem Total eruptions last 10 ka Eruptions VEI ≥ 3 a short time-and the tephra thickness is considered as a measure of the magnitude of the disturbance caused by the tephra burying the vegetation in the surrounding of the studied lake (regardless of the eruption magnitude, distance from the crater, wind direction, among other factors that could cause differences in the thickness of the deposited layer in a particular place). Furthermore, each tephra was coded as a quantitative explanatory variable and modeled as a simple exponential decay process (Lotter & Birks, 1993).
This model is a simple but robust equation (x −αt ), where x is the value for the ash (arbitrarily set to 100 by the authors), α is the decay coefficient equal to −0.5, and t is the sample depth (=time).
Also, the authors arbitrarily assigned a value before tephra deposition of 0. In this article, I preferred to describe the magnitude of the event of each tephra by giving the ash-value corresponding to the tephra thickness in centimeters. In doing so, I aim to describe the magnitude of the eruptions. Unlike the record of Lotter and Birks (1993), in sedimentary records from Patagonian the occurrence of multiple tephra is quite normal; therefore, the as value below a tephra layer corresponded to 0 only in some cases. The frequency of volcanic events was calculated as the sum of events over 1000 years and then was modeled with smoothspline method (spar = 0.7). Both variables, frequency and magnitude, were later used as an explanatory variable to constrain the pollen samples in a multivariate analysis (Redundant Analysis) assessing vegetation responses to volcanic disturbance regime.

| Reconstruction of the volcanic regime
The volcanic history and its disturbance regime were reconstructed from the number and thickness of tephra layers in the sediment of Lake Relem (Figure 2). The sediment core registers 39 tephra lay- with periods without volcanic disturbance around 7 and 5 ka. For the last 1.5 ka to the present, few tephra layers are recorded, but none deposited over 1.5 cm of tephra into the lake sediment.  plants traits (Antos & Zobel, 1986), and biological legacies (e.g., Dale et al., 2005). Moreover, those operating factors can vary in time and spatial extension from one eruptive event to another. In the following paragraph, I attempt to contextualize those patterns occurring around Lake Relem.

| Vegetation responses to volcanic events
Although Cyperaceae (a common element of the steppe grassland) showed the first ephemeral increase after the So-A eruption, Ephedra showed an unexpected increase dominating the early successional stage after the So-A eruption and not the expected Nothofagus species and Araucaria which are considered pioneer trees in the region ( Figure 5). Little is known about Ephedra's ecology and its palaeoecological significance in Patagonia, but in a study ~200 km southward from Lake Relem, it was found that Ephedra might have a nursery effect on Nothofagus and Austrocedrus after fire disturbance (e.g., Raffaele & Veblen, 1998) although Ephedra is not the most abundant species after disturbances. Modern pollen samples show that Ephedra averages 2.7% in Patagonia with maxima of 32.4% in some areas, but close to Lake Relem percentages are up to 20% (Paez et al., 2001). Ephedra's pollen grains have been known to disperse several kilometers far from the source area, but with low abundance (Maher, 1964). The pollen abundance of Ephedra before So-A eruption suggests a persistent presence but was always low; thus, the increase in pollen percentage and PAR following the So-A eruption should indicate a very local presence. Perhaps the species were suppressed or co-existing with other species before So-A eruption and the following increase as a result of the So-A eruption that facilitates its expansion under certain conditions. For instance, at Mallín Paso del Arco (Heusser et al., 1988), a record within the 2 m isopach zone of the So-A, a slight increase of Ephedra is seen around 3 ka, above a thick tephra layer at the core bottom. However, other sites do not show an increase in Ephedra related to tephra (Dickson et al., 2020;Fontana & Giesecke, 2017) and, to my knowledge, none of the palynological records in northern Patagonia show a significant increase of Ephedra after an eruption nor have been reported following recent eruptions.
Climatic condition may play a role in the environmental responses by determining the vegetation composition and structure, which in turn influence potential colonizer species and the rate of revegetation after disturbance. For example, before 4.5 ka when precipitation was low (Jenny et al., 2003;Lamy et al., 2001), grasses were more abundant (Figure 4). Thin tephra fall left some impact on the vegetation dominated by grasses (Figure 4b), but recovering fast enough to not register a peak in the RoC analysis In comparison, after the Taupo eruption in New Zealand, the vegetation returned to its original condition after about 120-250 years (Wilmshurst & McGlone, 1996). Later, when precipitation augmented in Patagonia, it promoted the development of Nothofagus species in the area. Then, the vegetation was less sensitive to small tephra deposition that might cause short-term effects-hardly detectable in the sediment-or was not disturbed significantly.
Future climate trends in northern Patagonia could be characterized by more frequent and intensive drought (Garreaud, 2019), Plant traits play also an important role in resisting tephra deposition and may determine the species performance. For example, Antos and Zobel (1986) found that herbs could not resist being buried by ~15 cm, while the early establishment was dominated by the shadow tolerant trees and shrubs under the undisturbed canopy. Regarding the So-A eruption, the early establishment of Ephedra was likely because of the taxon's special adaptation to poor soil, sprouting shoots, and the capacity to resist arid and cold weather conditions (Luebert & Pliscoff, 2006). Sprouting from roots seems to be a common mechanism of plants in response to buried disturbance by tephra fall. After the Taupo eruption (~1850 BP), the spores of Pteridium, a fern species with rhizome root system, were found increasing in several records in a wide area buried by tephra fall, thus suggesting the species spread after resisting the impacts of the eruption (Wilmshurst & McGlone, 1996). can be related to its fleshy-fruit and dispersal mechanism, principally through birds and rodents (Loera et al., 2015). Probably Ephedra was the major food supply for birds and rodents in a wide area devastated after So-A eruption. A similar mechanism of plant dispersal was suggested after the Taupo eruption, where the pollen of some fleshy-fruits taxa was found (Wilmshurst & McGlone, 1996), and in the last eruption of the Chaitén volcano, the species with fleshyfruits were also found playing an important role in the early vegetation establishment (Moreno- Gonzalez et al., 2019).
Trees are expected to be less affected by thin tephra fall, particularly Araucaria and Nothofagus species which are considered pioneer species after volcanic disturbances (Veblen, 1982). After the recent eruption of Puyehue, Nothofagus pumilio species was found F I G U R E 5 Successional patterns caused by the Sollipulli-Alpehue eruption. Primary y-axis shows pollen relative abundance (dotted curve) and secondary y-axis shows the pollen accumulation rate (continuous curve) of selected taxa. Small boxes at the top indicate successional phases. Pre-eruption is represented in green. Phase I (reddish area) indicates the "collapse" of the vegetation. Phase II (yellowish area), colonization and expansion into the disturbed area. Phase III, reorganization and recovery of original conditions. Note that periods of time are not exact and are used only with schematic purpose. resprouting in areas buried by tephra (Montiel et al., 2017). However, in the blast zone of the last eruption from Chaitén Volcano, none Nothofagus seedling were found 1 year after the eruption (Moreno- Gonzalez et al., 2019). In palynological records, the abundance of N. dombeyi type is always variable, and few studies have assessed directly the responses to volcanic events. Álvarez- Barra et al. (2020) demonstrated that well-developed forests were not affected in the long term by tephra fall smaller than 20 cm. However, after the large eruption of So-A, the pollen abundance of N. dombeyi-type was strongly affected (this study and Dickson et al., 2020). Indeed, the data presented here show that Nothofagus species did not behave as pioneer and expanded into the area only after 500 years in advanced successional stages following the decrease of Ephedra ( Figure 5, phase III).
Multivariate analyses indicated Araucaria was not affected by any of the thin tephra fall. It is largely documented that the thick bark, flexible branches and smooth leaves allow the species to resist moderate disturbance (Burns, 1991;Veblen et al., 2005).
Hence, it is possible that thin tephra fall deposited in the crown of the trees did not cause significant physical damage to the entire population. If chemical change to soil occurred after tephra deposition, physiological harm such as decreasing tree-ring growth might not last for a long time (Tognetti et al., 2012); therefore, it is unlikely to produce a significant change in pollen abundance from sedimentary records. Since Araucaria did not significantly change in pollen percentage and PAR after the So-A eruption (the largest eruption in the area for the last 9 ka), it is unlikely that volcanism affected negatively the distribution of the populations nor increased the forest fragmentation in areas distant from a volcanic source as it suggested by Bekessy et al. (2002). A recent pollen record from Lake Cilantro (Dickson et al., 2020) shows similar patterns to those observed in Lake Relem. The authors recorded neither changes in pollen relative abundance after small tephra layers nor changes in Open Access funding enabled and organized by Projekt DEAL.

CO N FLI C T O F I NTE R E S T
None conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Moreno- : Pollen and macro-charcoal analysis of Lake Relem sediment core, northern Patagonia. https://doi. org/10.1594/PANGA EA.923741.

O PE N R E S E A RCH BA D G E S
The data for this research are stored at the PANGAEA in the follow-