Carbon dioxide fluxes in Alpine grasslands at the Nivolet Plain, Gran Paradiso National Park, Italy 2017–2023 (2024)

Study site

The CZO@NIVOLET was installed in 2017 and is located within the boundaries of the Gran Paradiso National Park (GPNP). It is part of the Critical Zone Exploration Network (CZEN, https://www.czen.org/content/nivolet-czo), a global network investigating processes in the Critical Zone, which is defined as the dynamic living skin of the Earth that extends from the top of the vegetative canopy through the soil and down to fresh bedrock and the bottom of the groundwater¹⁷. This research site also belongs to both the European eLTER (https://elter-ri.eu/elter-ri) and ICOS ERIC (https://www.icos-cp.eu) Research Infrastructures (RI).

The GPNP was established in 1922 for the preservation of the Alpine ibex (Capra ibex) and the conservation of high-altitude mountain ecosystems. Encompassing an area of 720 km², the park features a wide range of ecosystems, including lower elevation Alpine woods, as well as high-altitude grasslands and Alpine tundra, rock cliffs, and glaciers above the treeline. The Nivolet Plain (Fig.1) is a glacial valley that ranges in elevation from approximately 2300 m a.s.l. in the northeast to around 2700 m a.s.l. in the southwest.

The underlying bedrock is composed of gneisses, dolostones and marbles from the Gran Paradiso Massif, as well as calcschists with serpentinites and metabasites from the Piedmont-Ligurian zone¹⁸.

Daily records of precipitation (mm), minimum, maximum, and mean temperature (°C) from the Lago Agnel weather station are freely available (CC BY-NC-SA 4.0) at Arpa Piemonte portal (https://www.arpa.piemonte.it/rischi_naturali/snippets_arpa_graphs/dati_giornalieri_meteo/?statid=PIE-001073-900-1996-10-10&param=P). According to such data, over the time span 2017–2023, the average daily minimum temperature from June to October was 6.11 °C, the average daily maximum temperature was 13.0 °C, and the average daily precipitation was 2.9 mm. During winter the soil is typically covered with a thick layer of snow.

The Nivolet Plain is home to Alpine natural grasslands that support a diverse array of species within the Caricion curvulae climax vegetation community¹⁹. Dominant species found in the grasslands include Carex curvula All., Alopecurus gerardi Vill., Gnaphalium supinum L., and Leontodon helveticus Mérat. In the investigation sites, also Geum montanum, Trifolium alpinum, Pulsatilla alpina, and Silene acaulis are commonly found. The plants in these high-altitude grasslands experience rapid development from late June to late October, with canopy heights reaching a maximum of 0.2 metres.

During summer, grasslands are grazed by both domestic and wild ungulates. Wild ungulates (ibex and chamois) are censused every year. In 2022, the population density in the GPNP was counted to be 2687 ibex individuals and 6346 chamois individuals (GPNP, unpublished data, see also²⁰), while there are no quantitative census data on roe deer, red deer, and wild boar. These latter, however, are typically found at lower altitudes than those considered in our study. Regarding domestic ungulates, from the beginning of July until mid-September, approximately 110 cows along with around 20 sheep, and goats are brought for grazing in this area¹⁹. Grazing is conducted in a controlled manner, with animals predominantly grazing in areas adjacent to the barn (located at 45°29'13.7“N 7°08'27.6“E) and throughout the lower regions of the Nivolet valley (see https://www.pastoralp.eu/homepage/ for more information).

The five measurement sites at the Nivolet Plain are located within the same hydrological basin, and each site has an area between ~ 500 to ~ 900 square metres. Three of the five selected sites are on the orographic left flank bordering the Nivolet Plain: one on carbonate rocks (site named CARB in the dataset, at 2750–2760 m a.s.l.) and two on glacial deposits (site named GLA, at 2740–2750 m a.s.l. and site named EC, at 2750–2760 m a.s.l.). One site lies on the orographic right flank of the Plain, on soils developed on gneiss (site named GNE, at about 2580–2600 m a.s.l.). One site is on alluvial soil at the Plain floor (site named AL, 2740–2750 m a.s.l.). The location of the five sites is shown in Fig.2. Mean coordinates of the five study sites are reported in Table1.

In 2020, D’Amico et al.²¹ published the soil types map of the Aosta Valley, encompassing the region that includes our study area. Besides, our research group conducted soil profile samplings in locations proximate to and with similar geological and geomorphological attributes of the five study sites. Some physical and chemical characteristics of soil profiles are briefly described here^22,23.

Since 2020, data on aboveground vegetation biomass at the EC site have become available. These values, calculated as averages from individual samples, are detailed in Table2.

Flux chamber measurements

In the summer of 2017, surveys were carried out to select measurement sites, assess instrumental setups, and determine the CO₂ flux ranges essential for laboratory calibration of portable flux chamber systems. This initial phase was characterised by few measurement campaigns. Starting from 2018, the frequency of these campaigns increased, establishing a regular schedule of measurements approximately every 10–15 days throughout the vegetative season. The final instrumentation setup is shown in Fig.3.

NEE and ER were measured using the non-steady state dynamic flux chamber method. The chamber, placed over the vegetated soil, isolates a volume of air where the concentration of CO₂ increases or decreases according to the dominant process at the soil-vegetation-atmosphere interface. In the presence of sunlight, if photosynthesis captures CO₂ faster than its release due to respiration, the CO₂ concentration inside the chamber decreases. If respiration is dominant over photosynthesis, the CO₂ concentration increases. CO₂ concentration inside the flux chamber is measured over a specific time interval; the flux is computed by interpolating the curve of CO₂ concentration versus time, as explained further in the text and discussed in detail here²⁴.

At each point, a stainless-steel collar was inserted for about 1 cm into the soil a few minutes before the measurement, assuring no leakage. Before placing the flux chamber, RGB (Red, Green, Blue) images were taken from a nadir perspective, aiming at monitoring the vegetation within the collar area. These images are freely available in the IGG-CNR-CZO community of the Zenodo repository²⁵.

The flux chamber was then placed on the collar to isolate a confined air volume (headspace). This created a closed system where the CO₂ concentration inside the chamber changes during the measurement because of CO₂ absorption by plants through photosynthesis and/or emission through respiration by autotrophs (i.e., plants) and heterotrophs (i.e., microbial communities in the soil).

Air from the headspace of the chamber was pumped at a constant flow rate of 3 l/min into an Infrared Gas Analyzer (IRGA, either model LI-840 or LI-850 CO₂/H₂O Analyzer; LI-COR Biosciences, Lincoln, NE, USA) through a 1.8-metres-long RILSAN® tubing. The sampled air was then reinjected into the chamber. The reinjection tube ended with a 0.40-metres-long coiled and pierced RILSAN® tube, which ensured good mixing of the reinjected air sample within the chamber. Before and after each measurement, the entire apparatus (including the chamber, tubing, and IRGA) was vented until the ambient CO₂ concentration was recorded and its concentration was stable for a few seconds.

The CO₂ concentration inside the flux chamber was measured for about 90 seconds. CO₂ concentration versus time was recorded at 1 Hz frequency using the custom Android app FluxManager2 (West Systems S.r.l., freely available on Google Play Store) which was installed on a palmtop computer connected to the instrument via Bluetooth. Upon completion of the measurements, a text file containing all the data, including meteo-climatic and environmental variables (see below), was generated in the internal memory of the Android device.

The concentration curve was interpolated linearly over a period of about 60 seconds to calculate the rate of change of CO₂ concentration over time (ppm s⁻¹). The interpolation was done using the custom FluxRevision software (West Systems S.r.l). The initial 10–15 seconds (cleaning time), and the final, potentially non-linear part of the curve were excluded from the interpolation.

Measurements were conducted at various times throughout the day, ranging from 10:00 to 18:00⁹, and covered different meteorological conditions, in order to capture the natural meteorological variability²⁶. For each measurement campaign and for each site, measurements were replicated at 15–20 different points within the site, randomly chosen to sample the small-scale flux variability. Previous analysis has shown that a minimum of 15 measurement points is generally sufficient to represent the spatial variability at these sites¹⁴.

Figure4 shows a typical measurement cycle, which included two consecutive measurements at each point: the first was performed under ambient light, using the transparent chamber to estimate NEE, while the second measurement was performed using the same chamber shaded with a cloth to estimate ER in the absence of photosynthesis. A similar procedure was applied in previous works on similar environments^26,27. This process was repeated at all points, requiring approximately 2 hours to cover an entire site. Notice that in 2020, owing to the restrictions imposed by the pandemics, the number of measurement campaigns had to be much reduced.

NEE and ER fluxes in μmolCO₂ m⁻² s⁻¹ were estimated from the slope of the linear regression of headspace CO₂ concentration over time (ppm s⁻¹) using a laboratory calibration curve that relates pre-determined CO₂ fluxes (in the range of fluxes expected in the field) with the corresponding measured slopes of the CO₂ vs time linear regression (see the section “Technical Validation”).

Mean values and variability of Net Ecosystem Exchange (NEE) and Ecosystem Respiration (ER) measured at site GNE (2017–2023) are illustrated in Fig.5, as an example of the data from one of the five sites. Part of the NEE data discussed here have been compared with the flux estimates provided by an eddy covariance tower located at the EC site, belonging to the FLUXNET network as ICOS-Associated ecosystemic station since 2022 (IT-Niv, ref to: https://meta.icos-cp.eu/resources/stations/ES_IT-Niv). The results of the comparison indicated that the site-average of the individual NEE point measurements at the EC site were consistent with the NEE estimates provided by the eddy covariance method for the same time and date¹⁵.

Meteo-climatic variables

The optimised version of the portable weather station, shown in Fig.3, was employed starting from 2018.

During the CO₂ flux measurements, FluxManager2 simultaneously recorded air temperature, atmospheric pressure, air relative humidity, solar irradiance, soil temperature, and soil volumetric water content (1 Hz acquisition).

Air relative humidity, air temperature, and solar irradiance were measured using LSI LASTEM thermohygrometers model DMA672.1 sheltered from direct solar radiation and LSI LASTEM pyranometers model DPA053A mounted on a portable tripod at a height of 1.5 metres above the ground²⁸. The atmospheric pressure was recorded using digital barometers placed inside the flux chambers.

Soil temperature and soil volumetric water content were measured using Pt100 thermometers and Delta-T SM150T soil moisture sensors at depths of approximately 10 cm for the soil temperature and in the range of 0–5 cm for the soil moisture. The measurements were taken at about 20 cm from the collar on undisturbed soil, specifically without removing the organic layer (layer O). To account for small scale variability of soil moisture, soil volumetric water content values were also taken inside the collar area before the measurements of CO₂ concentration to assess the moisture range (at least 3 measurements), then the probe was placed outside the collar, at a point where the soil water content was in the range of values measured inside the collar.

To ensure accuracy, these sensors were tested at CNR laboratories before and after each measurement season and calibrated in accredited laboratories every two years. The specifications for the sensors and probes are provided in Table3.

FluxManager2

The FluxManager2 Android application (West Systems S.r.l.) is installed on a palmtop computer provided with Bluetooth; it is used to manage the instrumentation, sensors and probes and for displaying and recording the data.

FluxManager2 Android app is freely available on Google Play Store.

FluxRevision

The FluxRevision software (West Systems S.r.l.) allows users to interpolate the CO₂ concentration curve and calculate their slope and R², using files created with FluxManager2. The software leaves the possibility to choose the linear interpolation interval.

FluxRevision is freely available for download from the West Systems website (https://www.westsystems.com/instruments/download/).