Preguntas específicas
Considerando distintas etapas de una sucesión forestal:
¿Cómo varían los caudales, dinámica de sedimentos y nutrientes esenciales durante eventos de precipitación?
¿Cuáles son los factores que modulan los procesos ecosistémicos durante eventos de precipitación?
Objetivos específicos
Analizar la respuesta de caudales durante eventos de precipitación Cuantificar el transporte de sedimentos ante eventos de precipitación
Cuantificar el flujo de nutrientes esenciales en distintos componentes del ecosistema, durante eventos de precipitación de distinta intensidad.
Relacionar morfometría de cuencas, atributos vegetacionales y propiedades físicas del suelo con respuesta de caudales, dinámica de sedimentos y nutrientes.
catchments of southern Chile
Cristián Frêne, Freddy Véliz, Fernando D. Alfaro & Juan J. Armesto
ABSTRACT
Rainfall modulates streamflow at watershed scale and many ecosystem processes in temperate forests. Runoff generation in small mountain catchments is characterized by a quick response to rainfall pulses that affects biogeochemical fluxes. In temperate climates, water erosion related to rainfall intensity is the most important factor in soil and nutrient losses. Catchment studies often focus on hourly or sub-hourly measurements of water fluxes, and weekly or monthly samples of rainfall and streamflow chemistry. We assessed the ability of forests at catchment scale to regulate hydrologic flows, sediment transport, and nutrient fluxes during rainfall events in different successional stages.
Comparisons among successional stages showed that catchments differed in their hydrological responses to high intensity rain events, with greater runoff in secondary forests (SF), compared to old-growth forests (OG) and scrubs (CH). Sediment loads showed high variability between and within successional stages, with SF catchments exporting 455 kg/ha, followed by OG with 91 kg/ha and CH with 14 kg/ha in 11 rainfall events measured. Rainfall nutrient content was enriched via throughfall. Suspended sediments, Nitrogen (TN) and Phosphorus (TP) concentrations in streamflow varied with rainfall intensity. CH catchments exported less nutrients (46 kg/ha TN and 7 kg/ha TP) than SF catchments (718 kg/ha TN and 107 kg/ha),
7 rainfall events. This research found additive effects of successional stage, related to attributtes such as vegetation structure and soil physical properties, as well as influences of catchment morphometry. In these complex ecosystems, hydrologic and biogeochemical fluxes are driven by multiple explanatory variables and interacting phenomena, including climate change and anthropogenic and natural distrubance history.
INTRODUCTION
The temporal variability of streamflow in response to rainfall modulates many ecosystem processes in wet temperate forests. For instance, sediment dynamics and bio-geochemical processes are affected by streamflow variation, which in turn play a significant role on ecosystem functions, such as hydrologic regulation, soil protection, and nutrient cycling (Becker 2005, Moraetis et al. 2010, Hrncir et al. 2010, Likens 2013).
Overland water flow or runoff generation in mountain catchments is one of the most complex and relevant ecosystem processes because it is highly variable in space and time, depending on synergistic combination of three main controlling factors: climate, soils and vegetation (Becker 2005, Likens 2013). The different combinations of these three factors determine the water balance of landscape units, along with sediment and nutrients fluxes that influence
characterized by a quick response to rainfall pulses (Becker 2005, Hrncir et al. 2010, Likens 2013) that affects biogeochemical fluxes.
Forest are complex ecosystems as a result not only of biological activities but also of many abiotic physical and chemical processes, who are essential drivers of material fluxes and its storage, so all organisms rely on them (Likens 2013, Smith et al. 2013, 2015). Water flow paths play a critical role in determining the hydrological and biogeochemical processes in forested watersheds (Bonell 1993, Cirmo & McDonnell 1997, Becker 2005, Monteith et al. 2006, Likens 2013). Spatial variability associated with topography and vegetation affects runoff generation and exerts strong controls on water and solute fluxes. This variability should be considered when monitoring and modelling watershed processes in response to anthropogenic or natural disturbances (Monteith et al. 2006, Likens 2013).
Seasonal streamflow in forested watersheds varies by orders of magnitude, whereas interannual streamflow varies by no more than twofold and is usually less (Likens 2013). Because the sediment and nutrients dynamics are so intimately tied to the hydrologic cycle, seasonal hydrologic events have corresponding effects on sediment transport (Holz et al. 2015) and chemical budgets (Likens 2013).
In humid temperate climates, water erosion connected with the intensity of rainfall is often the most important factor that induces soil loss (Geißler et al. 2013, Holz et al. 2015). Rainfall is a key driver of soil erosion by water as a product of kinetic energy and rainfall impact on soils
are used to describe rainfall erosivity, and both are derived from mass and velocity of raindrops, it is not entirely clear how different substrates transform this energy into sediment transport (Goebes et al. 2015, Holz et al. 2015).
The transport of soil particles to watercourses is mainly affected by forest type, soil physical properties, and topography (Tsukamoto 1989, Miura et al. 2002, Neary et al. 2009, Geißler et al. 2013, Holz et al. 2015, Frêne et al. 201X). Differences in forest successional stages within and among catchment in terms of vegetation structure and composition (Chen et al. 2016), as well as their capacity to modify soil physical properties (Frêne et al. 2018) are all relevant factors. In early stages of forest succession, where understory is less developed, erosion rates should be greater than in advanced stages of forest development because raindrops impact directly on the soil (Tsukamoto 1989, Geißler et al. 2013, Seitz et al. 2016). As forest succession progresses, erosion is expected to decrease due to the development of multiple plant layers in the vertical profile and the accumulation of litter, which together minimize the erosive effect of water (Miura et al. 2002, Neary et al. 2009, Geißler et al. 2013, Holz et al. 2015, Seitz et al. 2016, Chen et al. 2016).
Soil erosion alters ecosystem functioning by reducing soil organic carbon stocks, relocating nutrients and reducing resource supplies to plants, animals and microbes (Pimentel 2006, Goebes et al. 2015). Sediment exports are estimated to be between 1 and 5 tons per hectare per year in mountain areas of the world with natural vegetation (Pimentel 2006), making it a
transport affects soil nutrients by driving hydrological losses of particulate and dissolved nutrients such as Nitrogen (N), Phosphorus (P) and Carbon (C), as most organic matter accumulates on the litter layer or near the soil surface (Neary et al. 2009). The analysis of organic matter losses via hydrological flow paths in successional forests has deserved much attention in the literature, revealing that sediment transport by water or wind after forest disturbance are 1.3 to 5 times richer in organic matter than the remaining soil (Morgan 2005, Pimentel 2006).
Ecosystem monitoring
The watershed approach is a useful tool for assessing the impacts of natural and human disturbance on water, sediment and nutrient fluxes (Likens 2013). Hydrological monitoring in small headwater catchments provides the basis for examining complex interrelations of hydraulic processes that govern runoff (Becker 2005, Votrubova et al. 2017), along with sediment (Holz et al. 2015) and nutrient dynamics (Likens 2013). Temporary river hydrographs exhibit characteristic responses ranging from minutes to hours, such as experienced during first seasonal flush and storm events (Moraetis et al. 2010).
At present, most hydrochemical catchment studies have been based on hourly or sub-hourly measurements of water fluxes, and weekly or monthly samples of rainfall and streamflow chemistry. This stark mismatch in measurement time scales springs from the measurement technologies involved (Kirchner et al. 2004) and generates limitations and biases in
major hydrological and biological drivers of short-term variability in rainfall are not captured by conventional low-frequency monitoring programs (Neal et al. 2012).
Weekly or monthly monitoring of water chemistry have been widely used for assessing mass balances, or for detecting and measuring long-term trends in water quality (Kirchner et al. 2004, Moraetis et al. 2010, Neal et al. 2012). Relying on weekly or monthly sampling of chemical measurements restricts one’s scientific questions and worldviews to those that the data can address. The emergence of long-term, high-frequency chemical monitoring promises to open up new lines of inquiry that have previously been inaccessible (Neal et al. 2012). In particular, understanding process-linkages between catchment hydrology and streamwater chemistry requires chemical measurements in the time scale of hydrologic responses in small catchments, and the time scale of minutes or hours rather than weeks or months (Kirchner et al. 2004).
Forest hydrology studies have mostly been constrained to monthly or annual time scale, with no regard for processes occurring during single rainfall events, which could be major determinants of hydrological or biogeochemical fluxes (Moraetis et al. 2010, Neal et al. 2012). Seasonality of hydrological fluxes is relevant to understand the biogeochemical dynamics at the ecosystem scale (Likens 2013, Smith et al. 2013, 2015, Frêne et al. 2018), but the analysis of rainfall events is key to understand mechanisms of chemical and water balance (Moraetis et al. 2010, Neal et al. 2012, Goebes et al. 2015).
soils could enhance overland flow (Dörner et al. 2015). During the first seasonal flood a significant remobilization of accumulated debris occurs, a phenomenon described as the ‘first flush effect’ (Moraetis et al. 2010). These flush events transfer large quantities of sediments and nutrients from soil to streams (Moraetis et al. 2010, Likens 2013).
Up to now, low sampling frequency for water quality and quantity in space and time limits understanding of biochemical and geochemical hydrologic processes. Hydrologic and biogeochemical processes should be characterized by high-frequency monitoring and real-time observation of systems (Moraetis et al. 2010, Neal et al. 2012, Votrubova et al. 2017).
According to the classic successional hypotheses (Odum 1969, Vitousek & Reiners 1975, Gorham et al. 1979), the internal nutrient cycles are predicted to be more efficient in intermediate stages of forest succession, and therefore downstream nutrient losses will be the lowest in second-growth forests, with increasing values in advanced succession (old-growth forests) and early post-disturbance stages. The underlying mechanism postulated for explaining this pattern is that the higher rates of biomass accumulation in early stages of succession generate the capture, retention and storage of nutrients in living biomass, reaching maximum values in mid succession, with consequently reduced nutrient losses downstream (Gorham et al. 1979, Vitousek & White 1981).
The classic hypotheses have been challenged by data from studies in southern South American forests, which suggest the existence of tight hydrological controls on inorganic nutrient export
et al. 2007, 2008), together with large losses of elements in organic forms. Nevertheless, studies of forest hydrology and biogeochemistry in the southern hemisphere (Godoy et al. 2014) are characterized by low frequency monitoring, and have disregarded the dynamics of ecosystem processes in different stages of forest succession.
Rainfall variability and magnitude are important modulators of streamflow, nutrient fluxes and sediment transport, because of differences in energy to mobilize materials and driving water and nutrient export (Geißler et al. 2010, 2013, Goebes et al. 2015, Seitz et al. 2016). Most data about biogeochemical patterns during forest succession have been derived from northern hemisphere deciduous and conifer forests in temperate and boreal latitudes.
To reduce this geographical bias, we wished to assess the ability of southern South American forests to regulate hydrologic, sediment and nutrient fluxes downstream. We monitored small watersheds covered by native temperate forests, dominated by evergreen broad-leaved species, in different successional stages, addressing the following questions: how do vegetation and soils in different successional stages influence ecosystem processes?, what mechanisms modulate ecosystem processes during rainfall events?
The overall aim of this research was to assess the effects of rainfall events on streamflow regulation, sediment dynamics, and nutrient flows at catchment scale. Accordingly, we examined the processes controlling equilibrium and transient streamflow, sediment and nutrient dynamics during rain events.
streamflow and thereby enhance water storage will become more prominent as forest succession progresses, reaching maximum values in the old-growth stage. Hence, streamflow regulation during rainfall events will be stronger in advanced successional stages.
Regarding soil protection, we hypothesize that this ecosystem function will be most effective in early (scrub) and late stages (old-growth forest) of forest succession, reaching higher rates of erosion in the intermediate stages (secondary forests) of succession. This would be explained by the lower height (2-3 m above ground) and dense canopy cover, in the case of scrubs, and the multiple vegetational strata in the vertical profile of old-growth forest, in addition to the dense layer of litter present in both succesional stages, what as a whole should reduce kinetic energy of rainfall. Intermediate successional stages (secondary forests) would tend to less have soil protection due to the uniform and tall canopy height (> 15 m tall), poor vertical structure with low shrub cover, and low litter layer development.
Classic biogeochemical hypotheses (Odum 1969, Gorham et al. 1979) argue that old-growth forests are leaky (lose nutrients) compared to growing forests in early and mid-successional stages, which in contrast tend to retain nutrients (Perakis & Hedin 2002 for temperate forests, Brookshire et al. 2012 for tropical forest), but we hypothesize that there are important homeostatic mechanisms (Hedin et al. 1995, Salmon et al. 2001, Huygens et al. 2007, 2008) that allow ecosystem regulation of nutrients and sediment flows during rainfall events. These processes operate optimally in old-growth forests because of vertical structural complexity and
and multiple strata of old-growth forests, alongside with soil litter and soil physical characteristics should moderate nutrient exports in a similar way to early and mid successional stages.
METHODS
To address the question about the interaction of rainfall variability and forest successional stage, we assessed hourly changes in hydrologic parameters for 11 rainfall events of different magnitude (0.4-4.6 mm/hr), during the austral summer and fall initiation period (December 2013-April 2014, Table 1). Rainfall events were classified in two broad categories: low and high intensity, according to hydrologic responses observed, where low intensity events are those with less than 10 mm of total precipitation and rain intensity under 2 mm/hr.
The study was based on a chronosequence of forest stages (Walker et al. 2010) and followed the watershed approach (Likens 2013). For detailed information about the chronosequence and catchment descriptions see Frêne et al. (2018). In the Llancahue watershed (40°S), six headwater catchments were selected for this forest succession study, with two replicate catchments representing each of three contrasting forest successional stages (Figure 1) characteristic of southern temperate rainforests (Lara et al. 2014). Successional stages were pre-defined based on differences in plant species composition, structure (Table 1) and estimated ages since last disturbance: (1) early successional forest catchments were densely
some isolated patches of small trees, and originated from extensive slash and burn to open land for crops and pastures about 100 years ago and cattle grazing persisting until 20 years ago; (2) secondary forest catchments (SF) with a 20 m tall canopy dominated by evergreen
Nothofagus dombeyi, about 110 years old on average, derived from human-set fire and recolonization after disturbance; and (3) old-growth forest catchments (OG) characterized by the dominance of a diverse mix of 25 m tall, broad-leaved evergreen trees (Aextoxicon punctatum, Laureliopsis philippiana, Eucriphya cordifolia, Podocarpaceae and Myrtaceae species), with heterogeneous age structure and numerous trees about 350-400 years old. There was some evidence of small-scale disturbance due to limited selective logging.
Hydrologic flows
Rainfall inputs for the study area were estimated from data collected continuously by two rain gauges installed in open areas of the Llancahue watershed, connected to HOBO© data loggers (Onset Computer Corporation, Bourne, MA, USA). The gauges had a resolution of 0.25 mm and provided information about the number, duration, and volume of rainfall events. Runoff and base flows were recorded every 3 min with a HOBO© diver pressure transducer in V-notch weirs (60°), installed in the lower section of the six catchments, following standard procedures (Ward & Trimble 2004). To describe streamflow responses to rainfall events, we analysed a time series based on hourly records; each record is the mean value of 20 measurements per hour (sampling every 3 min).
Suspended solid transport was assessed with ISCO© (Teledyne Isco Inc., Lincoln, NE, USA) sampler devices installed in each of the six weirs selected for sampling. Hourly samples of suspended solids were obtained from the water course, making two samples of 500 mL each (every 30 min), obtaining a composite hourly sample of one liter. Samples were collected in plastic bottles (1000 mL), filtered in the laboratory with a vacuum pump, using fiberglass filters (ADVANTEC© GF75 47 mm diameter, 0.5 µm pore). Filters were initially washed with distilled water, dried at 60° C for 24 hours, and burned at 550° C for 20 min to eliminate impurities. The cleaned filters were weighed with a precision balance and used to filter the sediment samples. The filters with sediments were dried again at 60° C for 24 hours and weighed with precision balance to determine the total weight of suspended solids in stream water.
Water samples for N and P analysis were collected during 7 of the 11 rainfall events sampled. Rainfall inputs were obtained from rainwater samples with three gutters (0.225 m2 each) installed in an open area located next to the rain gauges. Throughfall samples were taken from three gutters (0.225 m2 each), installed under the forest canopy in each catchment (n=18 samples per rain event). Streamflow outputs of nutrients were assessed hourly with the same method for sediments previously described (ISCO© sampler devices were washed in laboratory previous to the rain events sampling). All samples were collected in bottles (60 mL Nalgene© HDPE) and frozen for subsequent chemical analyses in the laboratory.
CARY Institute (Millbrook, New York, USA), using a LACHAT (Quikchem 8500) auto sampler to obtain total N (TN) and total P (TP) with flow injection analysis method that comply with EPA, ISO, and DIN standards.
To assess possible statistical differences between and within successional stages, we performed a two-way ANOVA (factors: successional stage and catchments). In addition, we applied reaction norms and repeated measures ANOVA to test for differences among successional stages based on individual rainfall events.
RESULTS
Hydrologic flows
Surface runoff (Q) varied among rainfall events during the period of study, with major surface fluxes (> 50%) during high-intensity events, however in a large number of low-intensity events, base flows prevailed (90% of total flow) regardless of successional stage (Figure 2 A). Hydrograph comparisons among the three successional stages showed that, during high- intensity rainfall events, catchments presented different hydrological responses, displaying a single peak flow at the moment of high rain intensity, with SF catchments producing larger amounts of runoff water than other successional stages (Figures 2 A and S1). This hydrologic response was not observed under low intensity rainfall events for any successional stage (Figure S1).
and 4), but two-way ANOVA showed statistically significant differences among contrasting successional stages (p=0.0000691) as well as between catchments representing the same successional stage (p=0.02348), and no statistical differences in the interaction between these two factors (p=0.258; Table S1). One-way ANOVA and reaction norms showed no statistical differences between catchments representing the same successional stage (Figure 4).
Sediment transport
Suspended sediment concentration (SC) in streamflow varied with the intensity of rainfall events. In high intensity events, sediment concentrations are low at the beginning of the rainfall event and tend to increase with streamflow, reaching maximum concentrations at peak flow, or just before it, and dropping rapidly afterwards (Figure 2 B). The lowest sediment concentrations were found in CH and OG catchments (Figure 3). Two-way ANOVA showed statistical differences in sediment transport between catchments representing different successional stages (p=0.000952), but no statistical differences within catchments of the same successional stage (p=0.4597); the interaction between these two factors showed statistical differences (p=0.002; Table S1). One-way ANOVA and reaction norms showed statistical differences only between SF catchments (Figure 4).
Sediment loads showed high variability within and between successional stages (Table 1). Average sediment exports for the 11 rainfall events in SF catchments was 455 kg/ha (SD±440.8), followed by OG with 91 kg/ha (SD±91.4) and CH catchments with 14 kg/ha
Nutrient fluxes
Rainfall nutrient content was enriched via throughfall, as rainfall nutrient concentrations were very low in open field, ranging from 0.01 to 0.14 mg/L for TN and 0 to 0.01 mg/L for TP. Nutrient concentrations were higher under the forest canopy, ranging from 0.21 to 1.16 mg/L for TN and 0.03 to 0.13 mg/L for TP during the study period. TN imports via throughfall for the 7 rainfall events reached 1585.4 g/ha for CH catchments (SD±104.7), 4815.1 g/ha for SF catchments (SD±467.3) and 3347.9 g/ha for OG catchments (SD±1147.2, Table 2). TP imports via throughfall reached 175.2 g/ha for CH catchments (SD±11.6), 499.1 g/ha for SF catchments (SD±48.4) and 402.5 g/ha for OG catchments (SD±137.9, Table 2).
Nutrient concentrations in streamflow varied with the intensity of rainfall events in a similar way as sediment concentration (Figure 2). The lowest TN and TP concentrations were recorded in CH and OG catchments (Figures 2 and 3). Two-way ANOVA showed no statistical differences among successional stages (p=0.289 for TN and p=0,1671 for TP) and also within catchments of the same sucessional stage (p=0.7343 for TN and p=0.9861 for TP; Table S1); the interaction of the two factors showed statistical differences for TN (p=0.01048) but not for TP (p=0.2484; Table S1). One-way ANOVA and reaction norms showed statistical differences