Tracking canopy conductance and transpiration of CAM-plants Agave sisalana with carbonyl sulfide fluxes

Tracking canopy conductance and transpiration of CAM-plants Agave sisalana with carbonyl sulfide fluxes | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 13 May 2025 V1 Latest version Share on Tracking canopy conductance and transpiration of CAM-plants Agave sisalana with carbonyl sulfide fluxes Authors : Kukka-Maria Kohonen 0000-0001-9258-1225 [email protected] , Angelika Kübert , Lutz Merbold , Matti Räsänen 0000-0003-0994-5353 , Nina Buchmann 0000-0003-0826-2980 , Ivan Mammarella 0000-0002-8516-3356 , Petri Pellikka , and Timo Vesala 0000-0002-4852-7464 Authors Info & Affiliations https://doi.org/10.22541/au.174711610.06004106/v1 Published Agricultural and Forest Meteorology Version of record Peer review timeline 321 views 207 downloads Contents Abstract 1 Introduction 2 Materials and methods 3 Results 3.2 COS fluxes and their drivers 3.3 Canopy stomatal conductance and water fluxes 4 Discussion 5 Conclusions References Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Crassulacean acid metabolism (CAM) helps plants in arid regions to reduce water loss by opening their stomata and taking up carbon dioxide (CO 2 ) during nighttime. While gas exchange in CAM plants has been mainly studied under controlled laboratory conditions, only a few ecosystem scale studies exist. Moreover, carbonyl sulfide (COS) has been used as a tracer for stomatal conductance, transpiration and photosynthesis in C 3 and C 4 plants, but no studies on CAM ecosystems have yet been published. Here we present the first ecosystem scale measurements of COS fluxes over Agave sisalana (CAM plant) during the wet season in Kenya. The ecosystem was a consistent sink of COS, with higher uptake observed during nighttime (-11.5 pmol m −2 s −1 ) than during daytime (-5.6 pmol m −2 s −1 ). The magnitude of COS fluxes was comparable to non-growing season daytime fluxes reported for C 3 and C 4 plant dominated ecosystems. The soil was a small COS source (0.3 pmol m −2 s −1 ), with highest emissions under high radiation and temperature conditions. Using machine learning, we found that vapor pressure deficit, air temperature and soil water content were the most important drivers of nighttime ecosystem COS exchange (variable importance 0.25, 0.23 and 0.20, respectively), indicating the importance of stomatal limitation for COS fluxes. During daytime, air temperature, photosynthetically active radiation and soil temperature were the most important drivers (variable importances 0.19, 0.18 and 0.18, respectively). COS fluxes were further used to track canopy stomatal conductance and transpiration. Conductance values ranged from 0.03 mol m −2 s −1 during daytime to 0.06 mol m −2 s −1 during nighttime. Transpiration was thus higher during nighttime than during daytime, reflecting the CAM gas exchange strategy. Tracking canopy conductance and transpiration of CAM-plants Agave sisalana with carbonyl sulfide fluxes Kohonen, Kukka-Maaria11* Corresponding author: [email protected] 1,2 , Kübert, Angelika 2 , Merbold, Lutz 3,4 , Räsänen, Matti 5 , Buchmann, Nina 1 , Mammarella, Ivan 2 , Pellikka, Petri 6,7,8 and Vesala, Timo 2,5 1 ETH Zurich, Institute for Agricultural Sciences, Department of Environmental Systems Science, Zurich 8092, Switzerland 2 Institute for Atmospheric and Earth System Research/Physics, Faculty of Science, University of Helsinki, Helsinki, Finland 3 Integrative Agroecology Group, Research Division for Agroecology and Environment, Agroscope, Reckenholzstrasse 191, 8046 Zurich, Switzerland 4 Mazingira Centre, International Livestock Research Institute, Old Naivasha Road, 00100 Nairobi, Kenya 5 Institute for Atmospheric and Earth System Research/Forest Sciences, Faculty of Agriculture and Forestry, University of Helsinki, Helsinki, Finland 6 Department of Geosciences and Geography, University of Helsinki, P.O. Box 64, Helsinki 00014, Finland 7 State Key Laboratory of Information Engineering and Surveying, Mapping and Remote Sensing, Wuhan University, PR China 8 Wangari Maathai Institute for Environmental and Peace Studies, University of Nairobi, P.O. Box 29053, Kangemi 00625, Kenya Abstract Crassulacean acid metabolism (CAM) helps plants in arid regions to reduce water loss by opening their stomata and taking up carbon dioxide (CO 2 ) during nighttime. While gas exchange in CAM plants has been mainly studied under controlled laboratory conditions, only a few ecosystem scale studies exist. Moreover, carbonyl sulfide (COS) has been used as a tracer for stomatal conductance, transpiration and photosynthesis in C 3 and C 4 plants, but no studies on CAM ecosystems have yet been published. Here we present the first ecosystem scale measurements of COS fluxes over Agave sisalana (CAM plant) during the wet season in Kenya. The ecosystem was a consistent sink of COS, with higher uptake observed during nighttime (-11.5 pmol m − 2 s − 1 ) than during daytime (-5.6 pmol m − 2 s − 1 ). The magnitude of COS fluxes was comparable to non-growing season daytime fluxes reported for C 3 and C 4 plant dominated ecosystems. The soil was a small COS source (0.3 pmol m − 2 s − 1 ), with highest emissions under high radiation and temperature conditions. Using machine learning, we found that vapor pressure deficit, air temperature and soil water content were the most important drivers of nighttime ecosystem COS exchange (variable importance 0.25, 0.23 and 0.20, respectively), indicating the importance of stomatal limitation for COS fluxes. During daytime, air temperature, photosynthetically active radiation and soil temperature were the most important drivers (variable importances 0.19, 0.18 and 0.18, respectively). COS fluxes were further used to track canopy stomatal conductance and transpiration. Conductance values ranged from 0.03 mol m − 2 s − 1 during daytime to 0.06 mol m − 2 s − 1 during nighttime. Transpiration was thus higher during nighttime than during daytime, reflecting the CAM gas exchange strategy. Keywords : CAM, eddy covariance, COS, stomatal conductance, transpiration, flux measurements, ecosystem scale, Kenya 1 Introduction Plants with crassulacean acid metabolism (CAM) are well adapted to reduce water loss in arid environments by keeping their stomata closed during the day but opening them at night for carbon dioxide (CO 2 ) uptake. Because of this pronounced diel pattern, water use efficiency ( WUE ) of CAM plants can be three times higher than that of C 4 plants and six times higher than that of C 3 plants (Borland et al., 2009). In addition, the annual water demand of CAM plants is only one-quarter that of C 4 plants and one-sixth that of C 3 plants, making them highly drought tolerant (Borland et al., 2009; Yang et al., 2015). However, their biomass production is often lower than that of C 3 plants (Lüttge, 2004), although some Agave and Opuntia species have been found to reach similar or even higher productivity compared to C 3 plants (Nobel, 1996). The diel CAM gas exchange cycle consists of four phases as defined by Osmond (1978). During the night (phase I), CAM plants take up CO 2 , along with other gases, from the atmosphere. CO 2 is bound to malic acid by the enzyme phosphoenolpyruvate carboxylase (PEPC) and stored until light becomes available during the day. Then CO 2 is regenerated from malic acid and fixed by the enzyme ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) (phase III; Cushman, 2001). Phases II and IV of the CAM gas exchange cycle are the transient phases in the early morning and late afternoon, respectively, when CO 2 can also be assimilated directly in the Calvin cycle (Griffiths et al., 2007). Globally, CAM plants are estimated to make up over 6% of vascular plant species (Winter and Smith, 1996), but in arid regions, they can occupy up to 25% of the vegetative surface (Osmond, 1978). Currently, plants with CAM photosynthesis are not included in land surface models such as SiB4 (Kooijmans et al., 2021) or ORCHIDEE (Maignan et al., 2021), which only include plants with C 3 or C 4 photosynthesis. Although their global importance is limited, CAM plants may play a significant role in drylands which cover about 41% of Earth’s terrestrial surface (MEA 2005). Thus, while CAM plants might control regional carbon budgets, our knowledge of flux magnitudes of CAM-dominated ecosystems is scarce. One reason for their exclusion from land surface models is the lack of gas exchange measurements of CAM plants under field conditions. In the laboratory, gas exchange patterns of CAM plants have been measured, and the expected reverse diel pattern of CO 2 exchange compared to C 3 and C 4 plants has been found (Nobel and Valenzuela, 1987; Winter, 2019). Eddy covariance (EC) CO 2 flux measurements have been performed on tequila ( Agave tequilana ) (Owen et al., 2016), Opuntia stricta cactus (Jardim et al., 2023), pineapple ( Ananas comosus ) (San-José et al., 2007) and sisal ( Agave sisalana ) (Skogberg et al., 2025). Owen et al. (2016), San-José et al. (2007) and Skogberg et al. (2025) detected the CAM typical diel cycle with nighttime CO 2 uptake and daytime release or near-zero fluxes, whereas Jardim et al. (2023) only observed daytime CO 2 uptake. Moreover, Skogberg et al. (2025) found that the daytime CO 2 emission was greater during a dry season than during a wet season, while nighttime CO 2 uptake stayed at the same level of magnitude through both seasons. Previous studies on CAM ecosystems have found that the cumulative amount of radiation as well as temperature and vapor pressure deficit ( VPD ) during the day affected the CO 2 uptake of the following night (Owen et al., 2016). In addition, cold nights have been found to increase the nocturnal CO 2 uptake (Owen et al., 2016). However, previous ecosystem scale studies have focused solely on net ecosystem CO 2 exchange ( NEE ) and have not pursued partitioning NEE into respiration and gross primary production ( GPP ). Traditional NEE partitioning methods (Reichstein et al., 2005; Lasslop et al., 2010) cannot be directly implemented to CAM plants due to the reversed diel cycle compared to C 4 and C 3 plants as well as functional plasticity with non-CAM behavior. Moreover, nighttime CO 2 uptake does not necessarily equate to GPP , as not all carbon stored during the night may be used for photosynthesis during the next day. Hence, other approaches need to be employed for partitioning NEE fluxes of CAM ecosystems. Evapotranspiration ( ET ) is closely coupled with the CO 2 cycle through stomata, and this coupling is described by WUE (Keenan et al., 2013). WUE can be estimated from GPP and transpiration ( T ) as WUE = GPP/T , when it describes the photosynthetic carbon fixation and plant water loss (Stoy et al., 2019). Some studies, on the other hand, calculate WUE as WUE = NEE/ET , which provides a measure of the overall ecosystem carbon balance in relation to ecosystem water loss (Owen et al., 2016). ET can be calculated from the water vapor (H 2 O) fluxes measured by the EC method. However, partitioning ET into its components, evaporation ( E ) and transpiration, is challenging at ecosystem scale but important for the further improvement of land surface models (Stoy et al., 2019). A new development for ET partitioning was introduced by Zahn et al. (2022). This partitioning method, called conditional eddy covariance (CEC), uses raw EC data and distinguishes time periods when the fluctuation of both CO 2 and H 2 O is positive (emission), indicating soil respiration and E , while a negative fluctuation of CO 2 (uptake) simultaneously with a positive H 2 O fluctuation indicates photosynthesis and T . However, this method has not yet been widely applied or tested for terrestrial ecosystems, including CAM ecosystems. Carbonyl sulfide (COS) has attracted attention lately because it can be used as a proxy for stomatal conductance and thus for (gross) photosynthesis (Sandoval-Soto et al., 2005; Asaf et al., 2013) and transpiration (Wehr et al., 2017; Berkelhammer et al., 2020). COS is taken up by plants through stomata like CO 2 , until it reaches the chloroplast surface, where it is consumed in a hydrolysis reaction using the enzymes carbonic anhydrase (CA) and PEPC (Protoschill-Krebs and Kesselmeier, 1992; Wohlfahrt et al., 2012). To our knowledge, there are currently no studies focusing on the COS biochemistry of CAM plants. Since CA and PEPC are available in CAM plants similar to C 3 plants, the hydrolysis process is assumed to be similar. Complications of using COS as a proxy for GPP at ecosystem scale include unknown sinks and sources in the soil (Maseyk et al., 2014; Spielmann et al., 2020), the measurement precision of COS fluxes (Kohonen et al., 2020), and the uncertainty in estimating the leaf relative uptake ( LRU ), i.e., the ratio of COS and CO 2 uptake at leaf scale (Kooijmans et al., 2019; Yang et al., 2018; Sun et al., 2018; Kohonen et al., 2022). However, to date, field studies have mainly focused on C 3 (Asaf et al., 2013; Spielmann et al., 2019; Wehr et al., 2017; Vesala et al., 2022) and C 4 (Berkelhammer et al., 2020) plant-dominated ecosystems, and no COS flux measurements over CAM ecosystems have been published. In this study, we measured COS fluxes using the EC technique over a sisal ( Agave sisalana ) plantation in Kenya. The measurement period covered four weeks during the wet season in November-December 2019 and included 12 days of soil chamber flux measurements. Our aims were to 1) track the diel pattern of the CAM COS gas exchange, 2) identify its main environmental drivers, 3) quantify the canopy stomatal conductance of sisal using COS together with H 2 O flux measurements, 4) partition ET into its E and T components using the ecosystem scale COS, CO 2 and H 2 O flux measurements, and finally 5) calculate the WUE of sisal. 2 Materials and methods 2.1 Site description The measurements were carried out at an Agave sisalana Perrine plantation, Teita estate, in Mwatate in Taita-Taveta county, Kenya (3 ° 32’ S, 38 ° 24’E, 844 m asl) (Fig. S1). Sisal is cultivated for its fiber, traditionally used in making, e.g., ropes, carpets and baskets. From Teita estate, most of the fiber is transported overseas for production. The plantation covers an area of 129.5 km 2 , making it one of the largest in the world (Wachiye et al., 2021). Sisal has various growth stages, from recently planted up to 20 years old blocks. According to Vuorinne et al. (2021a, b), the biomass densities in the estate ranged from 0 to 46.7 Mg ha −1 according to the age classes (mean biomass 10.6 Mg ha −1 ). The flux measurements were made within a 2.6 ha field block growing sisal as a pure culture with spontaneous weeds coming in. The field block had a gentle 5° slope facing the east. Sisal plants were mature, 5 years old and approximately 1.1 m high (Skogberg et al. 2025). The estate has a yearly average temperature of 24 ° C and annual rainfall of approximately 600 mm. The soils are latosols, with low soil organic carbon content of around 1% (Pellikka et al., 2023). The area is characterized by dry as well as wet seasons, with a long wet season from March to May, followed by a long dry season from June until October, a short wet season from November to December, and a short dry season until February. The COS flux measurements were carried out during the short wet season from 23 November 2019 until 21 December 2019 (referred to as wet season in the following), while CO 2 flux, and meteorological measurements extended to the dry season until 26 January 2020. Herbicide was sprayed five days prior to the start of the measurements to prevent weeds with C3 or C4 photosynthesis growing in the EC footprint area. CO 2 fluxes from the same measurement campaign are reported in Skogberg et al. (2025). The times presented in this study are in local time (UTC+3), and nighttime is defined when photosynthetically active radiation PAR < 10 µ mol m − 2 s − 1 . Negative fluxes depict uptake by the ecosystem, while positive fluxes depict emissions. 2.2 Meteorological measurements Air temperature ( T air ) and relative humidity ( RH ) (Rotronic Instrument Corp., NY, USA) and PAR (LI-190R quantum sensor, LI-COR Inc., Nebraska, USA) were measured at 1.5 m above ground, approximately 0.5 m above the sisal leaves. Soil temperature ( T soil ; Pt100 thermocouples) was measured at two locations; one shaded by sisal and one without shading, both buried in topsoil at 2 cm depth. Soil water content ( SWC ; Theta Probe ML3, Delta-T Devices, Cambridge, UK) was measured in the topsoil (0–5 cm). All measurements were logged every minute and averaged every 30 min. Precipitation was measured with a tipping bucket rain gauge (Campbell ARG100, Campbell Scientific, Logan, Utah, USA) from 30.11.2019 onwards and logged every hour. Precipitation measurements 23.-30.11.2019 were taken as an average of two nearby weather stations in Maktau (run by University of Helsinki, approximately 30 km to the West) and Voi (run by Kenya Meteorological Department, approximately 30 km to the East). 2.3 Eddy covariance measurements and processing EC measurements were done at 2.6 m height using an ultrasonic anemometer (USA-1, Metek GmbH, Elmshorn, Germany) for measuring horizontal and vertical wind speeds (Fig. S1 a). COS, CO 2 and H 2 O mixing ratios were measured with a quantum cascade laser spectroscopy analyser (QCLS, Aerodyne Research Inc., Billerica, MA, USA). Additionally, CO 2 and H 2 O mixing ratios were measured with an infrared gas analyzer (IRGA, LI-7200, LI-COR Inc., Nebraska, USA). Inlet lines were 7 and 0.7 m long for the QCLS and IRGA, respectively, and the inner diameters were 4 mm. The inlet lines were heated to prevent water condensation in the tubing. All measurements were recorded at 10 Hz frequency. The fluxes were processed using EddyUH software (Mammarella et al., 2016), following the recommendations given in Kohonen et al. (2020). The processing included linear detrending, 2D coordinate rotation, despiking data (limits for subsequent datapoints being 50 ppm, 5 ppb and 10 mmol mol − 1 for CO 2 , COS and H 2 O, respectively) and spectral corrections for both low (Rannik, 1998) and high frequency (Aubinet et al., 2000) spectral losses. The limit of detection, estimated as suggested in Langford et al., (2015), was 5.8 pmol m -2 s -1 for COS fluxes, on average. Friction velocity ( u ∗ ) filtering was applied to remove periods with low atmospheric turbulence when u ∗ < 0.12 m s − 1 (Aubinet et al., 2000; Skogberg et al., 2025). Quality screening was applied so that maximum number of allowed spikes in each 30 min period was 100, the second wind rotation angle was less than 10°, and the mixing ratio was more than 200 ppt for COS and 300 ppm for CO 2 . After all quality screening and filtering, 44% of COS and 83% of CO 2 fluxes were left for analysis. The flux footprint covered only the sisal plantation (Fig. S1 b; footprint calculation after Kljun et al. (2015)). 2.4 Driver analysis and gap-filling The most important drivers of the fluxes were found and fluxes gap-filled using a model created with random forest (RF; RandomForestRegressor from Python package sklearn ). The models were trained using a randomly selected subset of the dataset (70%), the variables used as model features were CO 2 flux (for gap-filling COS and H 2 O fluxes only), H 2 O flux (for gap-filling COS and CO 2 fluxes), air and soil temperatures, VPD , PAR and SWC . The resulting models were validated against the test data set (coefficient of determination R 2 =0.62 and root mean square error RMSE=10.6 pmol m − 2 s − 1 for COS fluxes; R 2 =0.94 and RMSE=1.12 µ mol m − 2 s − 1 for CO 2 fluxes; and R 2 =0.95 and RMSE=0.396 mmol m -2 s -1 for H 2 O fluxes). To better interpret the model output, we used SHapley Additive exPlanations (SHAP) analysis, based on principles of game theory (Shapley, 1953). SHAP analysis (Python function TreeExplainer from package shap ) was done to study the contribution of each environmental driver to the model output, as well as to identify the temporal variation of the driver importance in more detail. Only measured data were used for models and driver analysis, while gap-filled flux data were used in time series analysis. Since CAM plants fix CO 2 into malic acid at nighttime, releasing it for actual photosynthetic CO 2 fixation during daytime (gross primary production, GPP ), we defined another variable to describe the amount of CO 2 fixed by the sisal plants during the night to be later used in daytime photosynthesis, i.e., the gross primary uptake ( GPU ). Ecosystem CO 2 fluxes were partitioned into GPU and R using the LI-7200 flux measurements from the (short) dry season, defined here as from 21 Dec to 26 Jan 2020 (Skogberg et al., 2025). The dry season daytime CO 2 flux was assumed to consist only of respiration (Skogberg et al., 2025) and used to train a random forest model for ecosystem respiration ( R eco ). The model was then applied for the whole dataset to model R eco and finally to get gross primary uptake as GPU = R eco - NEE . Based on this partitioning, we calculated sisal WUE = GPU/T . To better compare WUE with earlier studies, we also calculated an ecosystem scale WUE NET = NEE/ET . 2.5 Chamber measurements and flux calculation Chamber measurements were carried out at three different soil cover types: one with bare soil, one with non-CAM weeds growing on soil, and one with grass and old sisal roots on top of the soil (Fig. S1 c-e). One chamber head (height 40 cm, diameter 31.3 cm) was rotated for use on each soil cover type for both ambient light and darkened conditions. Stainless-steel chamber collars were installed one day before the first chamber measurement and left in place during the whole measurement campaign. The chamber head used in the study was cylindrical, made of acrylic plastic, and had a volume of 31 L. The chamber head had a small fan inside to mix the air, and air temperature inside the chamber was measured. Polytetrafluoroethylene (PTFE) tubing of 1/8” diameter was connected from the chamber to the QCLS for COS, CO 2 and H 2 O mixing ratio measurements. The sampling flow was set to 0.3 LPM, and each chamber was closed for 5 min. After closure at ambient light conditions, the chamber was flushed and covered for a darkened closure. For blank measurements (to ensure no emission or sink from the chamber itself), the collars were covered with PTFE foil and chamber measurements repeated as described above. The blank measurements showed no emissions of COS from the chamber. Chamber fluxes were calculated from the slope of the linear fit made to the concentration data collected during the first 3 min (for COS and CO 2 ) or 1 min (for H 2 O) of the chamber closure, before the concentration change was saturated. The measurements were omitted if the normalized RMSE of the fit was more than 0.3 or if R 2 < 0.8. 2.6 Canopy conductance Canopy conductance ( g c ) was calculated from COS fluxes ( FCOS ) according to Wohlfahrt et al., (2012) ( g c,COS ): \begin{equation} \begin{matrix}g_{c,COS}=\frac{-FCOS}{\chi_{COS,a}-\chi_{COS,i}}\ \#(1)\\ \end{matrix}\nonumber \\ \end{equation} where χ COS,a and χ COS,i are the dry mixing ratios of COS in the atmosphere and in the intercellular spaces, respectively. Assuming boundary layer and mesophyll conductances to be high (Kooijmans et al., 2019; Sun et al., 2024; Wehr et al., 2017), negligible soil fluxes, a very small χ COS,i , and a factor of 1.94 to account for the difference between COS and H 2 O conductances due to different molecular diffusivities (Seibt et al., 2010; Stimler et al., 2010), equation 1 reduces to \begin{equation} \begin{matrix}g_{c,COS}=\frac{-1.94\ FCOS}{\chi_{COS,a}}\ \#(2)\\ \end{matrix}\nonumber \\ \end{equation} Additionally, to have a comparison, canopy conductance was also calculated from ecosystem latent heat fluxes ( LE ) according to Wehr and Saleska (2021) ( g c,H 2 O ). In this method, both LE and sensible heat fluxes ( H ) were used together with the flux–gradient equations to determine canopy conductance. This method is less prone to biases than the widely used inversed Penman-Monteith equation (Monteith, 1965; Grace et al., 1995) due to the use of measured H . However, like the inversed Penman-Monteith equation, this method works properly only when transpiration is dominating over evaporation. In the case of CAM plants, this assumption is mostly met in the nighttime. However, without proper ET partitioning, it is impossible to know if the requirement is met. In this study, we report all g c,H2O values, regardless of high daytime evaporation. 2.7 ET partitioning COS-based transpiration ( T COS ) was calculated from the canopy conductance (Eq. 2) as in Berkelhammer et al. (2020) \begin{equation} \begin{matrix}T_{\text{COS}}=g_{c,COS}VPD\ \#(3)\\ \end{matrix}\nonumber \\ \end{equation} and COS-based evaporation ( E COS ) by subtracting T COS (Eq. 3) from the total evapotranspiration ( ET ) \begin{equation} \begin{matrix}E_{\text{COS}}=ET-T_{\text{COS}}.\ \#\left(4\right)\\ \end{matrix}\nonumber \\ \end{equation} Another ET partitioning method, the conditional eddy covariance (CEC) method proposed by Zahn et al. (2022), was also tested to validate the COS-based transpiration estimate. In the CEC method, raw EC data were divided into quadrants based on the turbulent fluctuations of vertical wind ( w’ ), CO 2 ( c’ ) and H 2 O ( q’ ) such that concurrent w’ >0, c’ >0 and q’ >0 indicated soil respiration and evaporation ( E CEC ), while simultaneous w’ >0, c’ 0 indicated CO 2 uptake ( GPU ) and transpiration ( T CEC ). In an earlier study, the CEC method performed best out of three different ET partitioning methods tested against independent estimates in a grassland in Kenya (Zahn et al., 2022) and was thus chosen for this study. 3 Results 3.1 Environmental conditions The measurement campaign from 23 Nov to 21 Dec 2019 took place during the short wet season (Fig. 1). The mean air temperature was 26.4 ° C during daytime and 21.2 ° C during nighttime. The maximum T air of 31.8 ° C was reached on 4 Dec in the afternoon, while the minimum T air of 17.9 ° C was measured in the same morning before sunrise. Soil temperature measured in a sun-exposed location ( T soil,exposed ) was always higher (median of 25.5 ° C) than the soil temperature at a shaded location ( T soil,shaded ; median 22.4 ° C). The maximum soil temperatures were 43.3 ° C on 19 Dec noon and 38.1 ° C on 25 Nov afternoon for the exposed and shaded locations, respectively. Minimum soil temperatures were 21.7 ° C on 23 Nov evening and 16.8 ° C on 25 Nov morning before sunrise for the exposed and shaded locations, respectively. Daytime maximum PAR varied from 1546 to 2374 µmol m -2 s -1 . Total precipitation from 30 Nov (when precipitation measurements on site started) to 21 Dec was 59 mm, while maximum daily precipitation was 15 mm on 1 Dec. SWC had its maximum value of 0.24 m 3 m − 3 in the beginning of the measurements and slowly decreased during the campaign to its minimum of 0.09 m 3 m − 3 on the last campaign day. A maximum VPD of 2.8 kPa was observed together with maximum air temperature on 4 Dec, while the minimum of 0.06 kPa was observed on 16 Dec. Median VPD during the campaign was 0.4 kPa. Local wind conditions closely followed the permanent regional east-to-west prevailing trade winds, typically coming from 60-90° directions. Maximum wind speed was 4.8 m s − 1 , with the median of 1.6 m s − 1 . Figure 1. Meteorological conditions during the measurement campaign 23 Nov – 21 Dec 2019: air (blue) and soil temperatures (sun exposed soil in orange, shaded soil in red) (a), photosynthetically active radiation ( PAR; b), daily rainfall (black bars) and soil water content ( SWC , blue line; c), and vapor pressure deficit ( VPD; d). 3.2 COS fluxes and their drivers The ecosystem showed net COS as well as CO 2 uptake during the measurement period 23 Nov- 21 Dec, with median nighttime EC fluxes of -11.5 pmol m − 2 s − 1 and -2.70 µ mol m − 2 s − 1 for COS and CO 2 , respectively, and daytime fluxes of -5.6 pmol m − 2 s − 1 and 0.46 µ mol m − 2 s − 1 for COS and CO 2 , respectively (Fig. 2). The COS fluxes were statistically different from zero ( p < 0.05) both during daytime and nighttime. The measurements thus showed the typical CAM gas exchange patterns with higher uptake at night than during the day for both COS and CO 2 . Moreover, the four CAM gas exchange phases (Osmond, 1978) were visible with high COS and CO 2 uptake at night (phase I, determined from sunset at 18:00 until sunrise at 6:00), increased uptake right after sunrise (phase II, at 6:00-9:00, determined from sunrise until PAR >1000 µ mol m − 2 s − 1 ), very low exchange rates (close to zero) in the mid-photoperiod (phase III, from approx. 9:00 to 15:00, determined from end of phase II until PAR < 1000 µ mol m − 2 s − 1 ), followed by low but steadily increasing uptake rates observed in the late afternoon (phase IV, 15:00 to 18:00, determined from end of phase III until sunset). The highest uptake (minimum flux) at hourly time resolution was observed at 7:00 (-15.9 pmol m − 2 s − 1 ) for COS and at 20:00 (-5.08 µ mol m − 2 s − 1 ) for CO 2 , while the lowest uptake (maximum flux) was observed at 10:00 for COS (1.6 pmol m − 2 s − 1 ) and at 13:00 CO 2 fluxes (1.51 µ mol m − 2 s − 1 ). Gross CO 2 uptake GPU , modelled with RF, was also higher during nighttime than daytime, but did not go to zero during the wet season, even during daytime (Fig. S3). Moreover, the nighttime GPU was of similar magnitude both during the dry and wet seasons. The mean fluxes over the whole measurement period were -8.6 pmol m − 2 s − 1 for COS and -1.16 µ mol m − 2 s − 1 for CO 2 . The ratio of COS to CO 2 deposition velocities, i.e., the median ecosystem relative uptake ( ERU ), was 1.7 in the night and 0.88 during daytime under high radiation ( PAR > 700 µ mol m − 2 s − 1 ). Opaque soil chamber measurements showed a small soil COS sink (-0.1 pmol m − 2 s − 1 on average), while with transparent chambers always COS emissions were measured (0.6 pmol m − 2 s − 1 on average). Overall, the soil was a small source of COS to the atmosphere (0.3 pmol m − 2 s − 1 ; Fig. S2). Moreover, the vegetated surface emitted more COS under radiation (1.1 pmol m − 2 s − 1 ) than the bare soil (0.4 pmol m − 2 s − 1 ). Soil COS emissions increased with radiation (from 0.1 pmol m − 2 s − 1 at PAR = 0 µmol m − 2 s − 1 to approx. 1 pmol m − 2 s − 1 at PAR > 1500 µmol m − 2 s − 1 ), and also with temperature (from 0.2 pmol m − 2 s − 1 at T air 40 ° C in the transparent chamber, and from -0.04 at T air 40 ° C in the dark chamber; Fig. 3f,h). Increasing T air , VPD and PAR decreased the COS uptake also at ecosystem scale (from -17 pmol m − 2 s − 1 at VPD < 0.2 kPa, T air < 22 ° C and PAR 2 kPa, T air > 29 ° C and PAR > 1300 µmol m − 2 s − 1 ) but increased soil COS emissions (Fig. 3). Occasional ecosystem scale emissions were observed at high T air and VPD during daytime, while during nighttime, no consistent emissions were observed. However, increasing T air during rain events increased COS uptake (Fig. S4). The most important drivers of COS fluxes at the ecosystem scale were T soil,exposed , PAR , T air , SWC and VPD according to the RF model (with variable importance of 0.23, 0.18, 0.16, 0.15, 0.15, respectively; Table S1). When the model was trained on nighttime data only, the most important variables were VPD , T air , SWC and T soil,exposed (variable importance of 0.25, 0.23, 0.20, 0.19, respectively), while for the daytime-trained model, most relevant variables were T air , SWC , PAR and T soil,shaded (variable importance of 0.19, 0.16, 0.18, 0.18, respectively). SHAP analysis revealed that low ( 24 ° C) limited daytime COS uptake (Fig. 4). Higher soil moisture ( SWC > 0.15 m 3 m − 3 ) increased both nighttime and daytime COS uptake (negative SHAP values), while very high SWC values ( SWC > 0.2 m 3 m − 3 ) during day limited COS uptake. VPD was mostly limiting COS uptake during nighttime and had a stronger net effect on the COS flux than during daytime (Fig. S5). A clear threshold value for VPD limitation was found at VPD > 0.6 kPa at nighttime (SHAP value turned positive, i.e., indicating limited COS uptake), while during the day, the limit was higher at VPD > 2.2 kPa. SWC increased COS uptake in the beginning of the campaign, but as SWC decreased, it started to limit COS uptake. Towards the end of the measurement period, after a rain event on 20 Dec, the SHAP value of SWC started to have negative values again, indicating increased COS uptake. Figure 2. Median diel variation of (a) COS (green), CO 2 (purple) and H 2 O (blue) fluxes and their 25 th and 75 th percentiles (shaded areas) and (b) canopy stomatal conductance calculated from COS ( g c,COS , green; Eq. 2) and H 2 O ( g c,H2O , blue; see Sect. 2.6 for details) flux measurements (left axis) and photosynthetic active radiation ( PAR , black, right axis). The grey area represents nighttime and vertical black lines indicate the four CAM gas exchange phases (roman numerals; see text for details). Figure 3. COS fluxes ( FCOS ) measured with eddy covariance (a-d) and soil chambers (e-h) against environmental drivers: soil water content ( SWC ; a, e), air temperature ( T air ; b, f), vapor pressure deficit ( VPD ; c, g) and photosynthetically active radiation ( PAR ; d, h). Data are separated into daytime (orange) and nighttime (blue) EC measurements, as well as between transparent daytime chamber (yellow) and daytime opaque chamber (purple) measurements. Data are binned to 7 or 8 equally sized bins, with each bin containing 35 datapoints for EC or 5 for chambers fluxes. Error bars represent the 75 th and 25 th percentiles, fit lines are based on second order polynomial functions. Figure 4. Shapley additive explanation (SHAP) values of T air , SWC and VPD based on random forest (RF) model for COS flux ( FCOS ) during nighttime (a-c) and daytime (d-f). Colors represent the modelled FCOS . Negative SHAP values indicate increased COS uptake or decreased COS emissions, while positive SHAP values indicate decreased net COS uptake or increased COS emissions. Note the different scaling of the x-axes for T air and VPD . 3.3 Canopy stomatal conductance and water fluxes Canopy stomatal conductance inferred from COS fluxes was lower than that from H 2 O fluxes during daytime (on average 0.03 and 0.24 mol m -2 s -1 from COS and H 2 O fluxes, respectively), but both conductances agreed well during nighttime (on average 0.06 and 0.08 mol m -2 s -1 from COS and H 2 O fluxes, respectively; Fig. 2b). Transpiration inferred from FCOS measurements and from the CEC method agreed well on a diel scale, although the CEC method showed always higher transpiration and lower evaporation than the COS method (Fig. 5). Nighttime T/ET was higher (0.66) than E/ET (0.34) for the CEC method, while the COS-based method showed higher nighttime E/ET (0.59) than T/ET (0.41). When VPD > 0.5 kPa, the nighttime T/ET fractions were highest at 0.45 and 0.81 for CEC and COS methods, respectively. During daytime, both methods showed that evaporation dominated over transpiration, with E/ET being 0.79 and 0.88 with CEC and COS-based methods, respectively, while T/ET was 0.21 and 0.12 with the CEC and COS-based methods, respectively. Evaporation calculated from soil chamber water fluxes was close to both ecosystem scale estimates, giving further proof for both ET partitioning methods (Fig. S6). Transpiration fraction T/ET increased with decreasing VPD during the day, while low VPD reduced transpiration in the night (Fig. 5c). On the other hand, evaporation fraction E/ET decreased with decreasing VPD during the day and increased during the night. WUE , calculated as WUE = GPU/T , was 69 µ mol CO 2 (mmol H 2 O) − 1 during nighttime and 3.9 µ mol CO 2 (mmol H 2 O) − 1 during daytime (Fig. S7), while the median over the whole measurement period was 23 µ mol CO 2 (mmol H 2 O) − 1 . To better compare with previous ecosystem scale studies, we also calculated WUE as WUE NET = NEE/ET . WUE NET had a median nighttime value of 7.3 µ mol CO 2 (mmol H 2 O) − 1 and daytime value of 0.1 µ mol CO 2 (mmol H 2 O) − 1 . Figure 5. Diel variation of (a) transpiration ( T/ET ) and (b) evaporation ( E/ET ) fractions, inferred from FCOS measurements (green) and from the conditional eddy covariance (CEC) method (blue; see Sect. 2.7 for details). The grey shaded area represents nighttime. Dependence of T/ET and E/ET fractions on vapor pressure deficit ( VPD ; c, d). Data in subplots c and d are binned to 15 equally sized bins. 4 Discussion 4.1 Temporal patterns of COS fluxes and their drivers The observed COS fluxes of the sisal plantation in Kenya during the short wet season were of the same magnitude as non-growing-season fluxes observed at a boreal forest (Vesala et al., 2022), a wheat field (Maseyk et al., 2014), and of a prairie and a maize field (Berkelhammer et al., 2020). However, the fluxes from the sisal plantation in this study were less than half of the magnitude of fluxes observed during the peak growing seasons in Vesala et al. (2022), Berkelhammer et al. (2020) and Maseyk et al. (2014), as well as those reported in a temperate mountain grassland, a Mediterranean savanna, a temperate beech forest and a soybean field (Spielmann et al., 2019). The low uptake of sisal compared to C 4 and C 3 plants measured in other studies could be explained by semi-arid conditions and low soil moisture, despite the wet season. Moreover, the ERU value found in this study (1.7 in the night, 0.88 in the day) was lower than that reported in previous studies (2.9±1.7 in Asaf et al., 2013 and 3.3±1.1 in Maseyk et al., 2014), indicating lower relative COS uptake compared to CO 2 uptake. This may indicate differences in the biochemical COS uptake in CAM plants compared to C 4 and C 3 plants. The soil COS fluxes were of similar magnitude as those observed in an agricultural field (Maseyk et al., 2014) and in a grassland (Berkelhammer et al., 2014). Increasing soil COS emissions under high temperatures and radiation in agricultural land have also previously been observed by Maseyk et al. (2014) and Kitz et al. (2020). In this study, the diel cycle of the COS flux closely followed that of the CO 2 flux. Both revealed the typical CAM gas exchange pattern (Osmond, 1978), with highest uptake observed in the night during phase I, a strong uptake peak right after sunrise in the beginning of phase II, reduced uptake (or net emission in the case of CO 2 flux) in phase III during midday, and a transitioning phase in the late afternoon in phase IV before sunset. In many species, CAM photosynthesis occurs together with C 3 photosynthesis, which only occurs during daytime, demonstrating a large degree of functional plasticity in response to water stress (Cushman, 2001; Winter, 2019). However, Agave species have generally been found to be constitutive CAM, i.e., almost completely CAM plants (Winter, 2019), although they can also respond to high water availability in the wet season with added daytime C 3 photosynthesis (Skogberg et al., 2025). Since the ecosystem scale COS fluxes were more negative than the simultaneous daytime soil COS fluxes, our data indicated also daytime stomatal COS uptake during the wet season. This was further supported by the GPU estimate that remained at the same level during wet and dry seasons during nighttime (Skogberg et al., 2025), while the wet season daytime GPU was higher than dry season daytime GPU , which was very small. This clearly showed that the sisal stomata remained open during daytime in the wet season, maintaining both C 3 and CAM photosynthesis. Air temperature was a more important driver than the soil temperature, which suggested a stomatal limitation on COS fluxes because stomata are more influenced by changes in atmospheric conditions than in the soil (Göbel et al., 2019). VPD was an important driver both during daytime and nighttime, again indicating stomatal limitation (Kooijmans et al., 2019). The SHAP analysis implied that stomata were closed during the day when T air was high, and that stomata were closed during night with low T air and limited COS uptake. In contrast, Owen et al. (2016) found increased CO 2 uptake during cold nights for a different Agave species, Agave tequilana , pointing towards higher stomatal activity. Low soil moisture was limiting both daytime and nighttime COS uptake. This limitation is due to low CA activity under dry soil conditions (Kevrešan et al., 1997), restraining the hydrolysis reaction and thus COS uptake. High soil moisture in our study, on the other hand, limited daytime COS uptake, likely due to heavy rain events temporarily reducing COS uptake through reduced stomatal conductance. 4.2 COS as proxy for canopy conductance and transpiration Canopy conductances inferred from COS and H 2 O fluxes followed opposite diel cycles, with g c,COS having its maximum values at night but g c,H2O having its maximum during the day. Transpiration fraction was higher during night than during day, as expected for a CAM ecosystem. The daytime difference in g c,COS and g c,H2O was likely due to high E/ET fractions, since assumptions underlying g c,H2O rely on ET to be dominated by T (Wehr & Saleska, 2021). However, since both methods agreed well during the night when T/ET fraction was at its highest, the COS-based conductance estimate seems reasonable. Moreover, the magnitude of the nighttime g c was similar, although slightly lower than observed in Wehr et al. (2017) for a temperate deciduous forest during daytime (0.1 mol m -2 s -1 ) and slightly higher than (modelled) canopy conductance values for a pineapple field in the wet season (San-José et al., 2007). The COS-based transpiration estimate followed a similar diel cycle as the CEC-based estimate but was consistently lower than the CEC-based transpiration. Daytime transpiration was low with both methods. However, as our GPU estimate showed gross CO 2 uptake also during the day (Fig. S3), it is likely that the very low COS-based transpiration estimate was underestimated, especially during the day. Moreover, atmospheric T air and VPD, used in the COS-based calculation of g c,COS and T COS , were probably different from leaf temperature and VPD experienced by the sisal leaves. Leaf temperatures are typically higher than T air (Schulze et al., 2019), and thus COS-based g c,COS and T COS likely underestimated (Vesala et al., 2005). Moreover, since there are no studies on mesophyll COS conductance in CAM plants, we assumed it to be similar as for C 3 plants, i.e., higher than stomatal conductance (Kooijmans et al., 2019; Sun et al., 2024) and was thus ignored in the total conductance calculation. Both ET partitioning methods showed very low transpiration during nighttime at high humidity, as expected. If mesophyll conductance would limit COS exchange, the T COS estimate would also decrease, while the COS-independent T CEC estimate would not. However, we did not observe higher limitations in the COS-based T COS estimate compared to the CEC-based T CEC even in dry conditions. This indicated that the mesophyll conductance did not start to limit COS uptake even in dry conditions, unlike observed in Wohlfahrt et al. (2018) and Sun et al. (2024). This further highlighted the higher drought resilience of sisal compared to the Mediterranean pine forest and coast live oak studied by Wohlfahrt et al. (2018) and Sun et al. (2024), respectively. 4.3 Water use of sisal Nighttime WUE of 68 µ mol CO 2 (mmol H 2 O) − 1 was higher than WUE values reported for CAM plants in earlier (laboratory) studies (4–10 µ mol CO 2 (mmol H 2 O) − 1 in Borland et al., 2009) and 6–30 µ mol CO 2 (mmol H 2 O) − 1 in Lüttge, 2004). However, average WUE calculated over 24 hours (35 µ mol CO 2 (mmol H 2 O) − 1 ) was very similar to what was reported in a laboratory study in Ellen & Ferrari (1997) (up to 42 µ mol CO 2 (mmol H 2 O) − 1 ). WUE changes with the CAM phase, with lowest WUE observed in phase IV (Lüttge, 2004), as also observed in our study. Daytime WUE of 3.9 µ mol CO 2 (mmol H 2 O) − 1 , on the other hand, was comparable to WUE of C 3 plants (Borland et al., 2009), again highlighting the photosynthetic plasticity of sisal during the wet season and good water availability. The ecosystem WUE NET (7.3 µmol CO 2 (mmol H 2 O) −1 ), on the other hand, was very similar to the 6.9 µ mol CO 2 (mmol H 2 O) − 1 reported by Owen et al. (2016) for Agave tequilana during nighttime in the wet season. Due to the high WUE observed in CAM plants, they are essential in mitigating climate change in arid regions (Yang et al., 2015; Borland et al., 2015). 5 Conclusions COS fluxes measured over A. sisalana showed the CAM-typical diel gas exchange pattern with higher uptake observed at night than during the day. However, the uptake continued also under radiation, indicating the combination of CAM and C 3 photosynthesis of the sisal plantation during the wet season. Besides radiation, air temperature, SWC and VPD were the most important environmental drivers of COS dynamics, showing that ecosystem COS fluxes were mostly controlled by stomatal closure. Soil COS fluxes were small but increased under high radiation and temperature. Canopy stomatal conductance and transpiration inferred from COS fluxes reflected canopy performance better than calculated based on H 2 O fluxes. This is due to daytime ET being dominated by evaporation, not accounted for in the H 2 O flux-based canopy conductance model, while COS fluxes directly track stomatal conductance and transpiration. Moreover, the COS-based transpiration estimate compared well with an independent approach to estimate T based on raw CO 2 and H 2 O EC data. Our results show that COS fluxes provide reliable estimates for canopy stomatal conductance as well as for ET partitioning, also for CAM plant-dominated ecosystems. Acknowledgements We want to thank Sami Haapanala and the staff at the Taita Research Station of the University of Helsinki, especially Mwadime Mjomba for constructing the base for the measurement station as well as for helping with station maintenance and set up, and Muhia Gicheru and Ambrose Nyga for setting up the measurement station. We also want to thank the very helpful staff at the International Livestock Research Institute (ILRI), especially George Wanyama, for assisting with shipping of instruments and ordering gas bottles. We would like to thank the Teita sisal estate for making it possible for us to conduct the study and for supporting us with logistics and electricity. Without their interest and cooperation, our study would not have been possible. We thank Mikko Skogberg for chamber flux calculations. Research permission from NACOSTI (no. P/18/97336/26355) is acknowledged. Funding information The setup of the measurement campaign was funded by the Research Council of Finland for SMARTLAND (Environmental sensing of ecosystem services for developing a climate-smart landscape framework to improve food security in East Africa, decision no. 318645) for TV and PP. The research was funded by University of Helsinki (ICOS-Finland). LM acknowledges funding received from the European Union’s Horizon Europe Programme (grant agreement number 101058525) for the project ”Knowledge and climate services from an African observation and Data research Infrastructure (KADI)”. Competing interests The authors declare that they have no compering interests. Author contributions KMK, TV, and PP designed the study and contributed to setting up the measurements. LM and MR contributed to the measurement setup and helped with logistics. KMK performed the measurements and processed and analyzed the flux data. All authors contributed by commenting on the final study design, results, and the manuscript. KMK wrote the manuscript with contributions from all co-authors. Data availability All data will be made openly accessible upon publication. References 1. Asaf, D., E. Rotenberg, F. Tatarinov, U. Dicken, S. A. Montzka, and D. Yakir (2013): Ecosystem photosynthesis inferred from measurements of carbonyl sulphide flux, Nature Geoscience 6 , 186–190. Aubinet, M., A. Grelle, A. Ibrom, Ü. Rannik, J. Moncrieff, T. Foken, A. S. Kowalski, P. H. Martin, P. Berbigier, C. Bernhofer, R. Clement, J. Elbers, A. Granier, T. Grünwald, K. Morgenstern, K. Pilegaard, C. Rebmann, W. Snijders, R. Valentini, and T. Vesala (2000): “Estimates of the annual net carbon and water exchange of forests: the EUROFLUX methodology”, Advances in ecological research , vol. 30, pp. 113–175. Berkelhammer, M., B. Alsip, R. Matamala, D. Cook, M.E. Whelan, E. Joo, C. Bernacchi, J. Miller, and T. Meyers (2020): Seasonal evolution of canopy stomatal conductance for a prairie and maize field in the midwestern United States from continuous carbonyl sulfide fluxes, Geophysical Research Letters 47 .6, e2019GL085652. Berkelhammer, M., D. Asaf, C. Still, S. Montzka, D. Noone, M. Gupta, R. Provencal, H. Chen, and D. Yakir (2014): Constraining surface carbon fluxes using in situ measurements of carbonyl sulfide and carbon dioxide, Global Biogeochemical Cycles 28 .2, 161–179. Borland, A. M., H. Griffiths, J. Hartwell, and J. A. C. Smith (2009): Exploiting the potential of plants with crassulacean acid metabolism for bioenergy production on marginal lands, Journal of Experimental Botany 60 .10, 2879–2896. Borland, A. M., S. D. Wullschleger, D. J. Weston, J. Hartwell, G. A. Tuskan, X. Yang, and J. C. Cushman (2015): Climate-resilient agroforestry: physiological responses to climate change and engineering of crassulacean acid metabolism Crossref Google Scholar 2. Cushman, J. C. (2001): Crassulacean acid metabolism. A plastic photosynthetic adaptation to arid environments, Plant physiology 127 .4, 1439–1448. Göbel, L., H. Coners, D. Hertel, S. Willinghöfer, and C. Leuschner (2019): The role of low soil temperature for photosynthesis and stomatal conductance of three graminoids from different elevations, Frontiers in plant science 10 , 330. Grace, J., J. Lloyd, and J. McIntyre (1995): Fluxes of carbon dioxide and water vapour over an undisturbed tropical forest in south-west Amazonia, Glob. Change Biol. , 1, 1–12. Griffiths, H., A. B. Cousins, M. R. Badger, and S. von Caemmerer (2007): Discrimination in the dark. Resolving the interplay between metabolic and physical constraints to phospho enol pyruvate carboxylase activity during the crassulacean acid metabolism cycle, Plant Physiology 143 .2, 1055–1067. Jardim, A. M. D. R. F., J. E. F. de Morais, L. S. B. de Souza, F. R. Marin, M. S. B. de Moura, L. P. C. Morellato, A. A. Montenegro, J. P. H. B. Ometto, J. L. M. P. de Lima, J. C. B. Dubeux Júnior & T. G. F. da Silva (2023): Sink or carbon source? how the Opuntia cactus agroecosystem interacts in the use of carbon, nutrients and radiation in the Brazilian semi-arid region, Journal of Hydrology 625, 130121. Keenan, T. F., D. Y. Hollinger, G. Bohrer, D. Dragoni, J. W. Munger, H. P. Schmid, and A. D. Richardson (2013): Increase in forest water-use efficiency as atmospheric carbon dioxide concentrations rise, Nature 499 .7458, 324–327. Kevrešan, S., B. Krstić, M. Popović, and L. Kovačev (1997): Biochemical changes in sugar beet lines in dependence on soil moisture, Biologia Plantarum , 39 (2), 289–293. Kitz, F., F. M. Spielmann, A. Hammerle, O. Kolle, M. Migliavacca, G. Moreno, A. Ibrom, D. Krasnov, S. M. Noe, and G. Wohlfahrt (2020). Soil COS exchange: a comparison of three European ecosystems. Global Biogeochemical Cycles , 34 (4), e2019GB006202. https://doi.org/10.1029/2019GB006202. Kljun, N., P. Calanca, M. W. Rotach, and H. P. Schmid (2015): A simple two-dimensional parameterisation for Flux Footprint Prediction (FFP), Geoscientific Model Development 8.11, 3695–3713. Kohonen, K-M., P. Kolari, L.M.J. Kooijmans, H. Chen, U. Seibt, W. Sun, and I. Mammarella (2020): Towards standardized processing of eddy covariance flux measurements of carbonyl sulfide, Atmospheric Measurement Techniques 13 .7, 3957–3975. Kohonen, K-M., D. Roderick, G. Tramontana, A. Mauranen, P. Kolari, L. M. J. Kooijmans, D. Papale, T. Vesala, and I. Mammarella (2022): Intercomparison of methods to estimate gross primary production based on CO 2 and COS flux measurements, Biogeosciences 19.17, 4067–4088. Kooijmans, L. M.J., A. Cho, J. Ma, A. Kaushik, K.D. Haynes, I. Baker, I.T. Luijkx, M. Groenink, W. Peters, J. B. Miller, J. A. Berry, J. Ogée, L. K. Meredith, W. Sun, K-M. Kohonen, T. Vesala, I. Mammarella, H. Chen, F. M. Spielmann, G. Wohlfahrt, M. Berkelhammer, M. E. Whelan, K. Maseyk, U. Seibt, R. Commane, R. Wehr, and M. Krol (2021): Evaluation of carbonyl sulfide biosphere exchange in the Simple Biosphere Model (SiB4). Biogeosciences, 18 (24), 6547-6565, https://doi.org/10.5194/bg-18-6547-2021. Kooijmans, L.M.J., W. Sun, J. Aalto, K-M. Erkkilä, K. Maseyk, U. Seibt, T. Vesala, I. Mammarella, and H. Chen (2019): Influences of light and humidity on carbonyl sulfide-based estimates of photosynthesis, Proceedings of the National Academy of Sciences 116 .7, 2470–2475. Langford, B., W. Acton, C. Ammann, A. Valach, and E. Nemitz (2015): Eddy-covariance data with low signal-to-noise ratio: time-lag determination, uncertainties and limit of detection, Atmospheric Measurement Techniques 8 , 4197–4213. Lasslop, G., M. Reichstein, D. Papale, A. Richardson, A. Arneth, A. Barr, P. Stoy and G. Wohlfahrt (2010): Separation of net ecosystem exchange into assimilation and respiration using a light response curve approach: critical issues and global evaluation, Global Change Biology 16 , 187–208. Lüttge, U. (2004): Ecophysiology of crassulacean acid metabolism (CAM), Annals of botany , 93 (6), 629-652. Maignan, F., C. Abadie, M. Remaud, L. M. J. Kooijmans, K.-M. Kohonen, R. Commane, Crossref Google Scholar 3. Mammarella, I., O. Peltola, A. Nordbo, L. Järvi, and Ü. Rannik (2016): Quantifying the uncertainty of eddy covariance fluxes due to the use of different software packages and combinations of processing steps in two contrasting ecosystems, Atmospheric Measurement Techniques 9 .10, 4915–4933. Maseyk, K., J. A. Berry, D. Billesbach, J. E. Campbell, M. S. Torn, M. Zahniser, and U. Seibt (2014): Sources and sinks of carbonyl sulfide in an agricultural field in the Southern Great Plains, Proceedings of the National Academy of Sciences 111.25, 9064– 9069. Millennium Ecosystem Assessment (2005): Ecosystems and Human Well-being: Synthesis. Island Press, Washington, DC. Monteith, J. (1965): Evaporation and environment, Symp. Soc. Exp. Biol. , 19, 205–234. Nobel, P.S. (1996): “High Productivity of Certain Agronomic CAM Species”, Crassulacean Acid Metabolism: Biochemistry, Ecophysiology and Evolution , ed. by Klaus Winter and J. Andrew C. Smith, Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 255–265, isbn: 978-3-642-79060-7. Nobel, P. S. and A. G. Valenzuela (1987): Environmental responses and productivity of the CAM plant, Agave tequilana, Agricultural and Forest Meteorology 39 .4, 319– 334, issn: 0168-1923. Osmond, C. B. (1978): Crassulacean acid metabolism: a curiosity in context, Annual review of plant physiology 29 .1, 379–414. Owen, N. A., Ó. N. Choncubhair, J. Males, J. I. del Real Laborde, R. Rubio-Cortés, H. Griffiths, and G. Lanigan (2016): Eddy covariance captures four-phase crassulacean acid metabolism (CAM) gas exchange signature in Agave, Plant, Cell & Environment 39 .2, 295–309. pce.12610. Protoschill-Krebs, G. and J. Kesselmeier (1992): Enzymatic pathways for the consumption of carbonyl sulphide (COS) by higher plants, Botanica Acta 105.3, 206–212. Pellikka, P., M. Luotamo, N. Sädekoski, J. Hietanen, I. Vuorinne, M. Räsänen, J. Heiskanen, M. Siljander, K. Karhu & A. Klami (2023): Tropical altitudinal gradient soil organic carbon and nitrogen estimation using Specim IQ portable imaging spectrometer. Science of the Total Environment 883, 16367. Rannik, Ü. (1998): On the surface layer similarity at a complex forest site, Journal of Geophysical Research: Atmospheres 103 .D8, 8685–8697. Reichstein, M., E. Falge, D. Baldocchi, D. Papale, M. Aubinet, P. Berbigier, C. Bernhofer, N. Buchmann, T. Gilmanov, A. Granier, T. Grünwald, K. Havránková, H. Ilvesniemi, D. Janous, A. Knohl, T. Laurila, A. Lohila, D. Loustau, G. Matteucci, T. Meyers, F. Miglietta, J.M. Ourcival, J. Pumpanen, S. Rambal, E. Rotenberg, M. Sanz, J. Tenhunen, G. Seufert, F. Vaccari, T. Vesala, D. Yakir, and R. Valentini (2005): On the separation of net ecosystem exchange into assimilation and ecosystem respiration: Review and improved algorithm, Global Change Biology 11 , 1424–1439. San-José, J., R. Montes, and N. Nikonova (2007): Diurnal patterns of carbon dioxide, water vapour, and energy fluxes in pineapple [Ananas comosus (L.) Merr. cv. Red Spanish] field using eddy covariance, Photosynthetica 45 , 370–384. Sandoval-Soto, L., M. Stanimirov, M. Von Hobe, V. Schmitt, J. Valdes, A. Wild, and J. Kesselmeier (2005): Global uptake of carbonyl sulfide (COS) by terrestrial vegetation: Estimates corrected by deposition velocities normalized to the uptake of carbon dioxide (CO 2 ), Biogeosciences 2 , 125–132. Schulze, E.D., E. Beck, N. Buchmann, S. Clemens, K. Müller-Hohenstein, M. Scherer-Lorenzen (2019): Plant Ecology. 2 nd edition. Springer. pp 928. Seibt, U., J. Kesselmeier, L. Sandoval-Soto, U. Kuhn, and J. A. Berry (2010): A kinetic analysis of leaf uptake of COS and its relation to transpiration, photosynthesis and carbon isotope fractionation, Biogeosciences 7 .1, 333–341. Shapley, L. S. (1953): Stochastic games, Proceedings of the national academy of sciences 39 .10, 1095–1100. Skogberg, M., K-M. Kohonen, A. Lohila, L. Merbold, M. Räsanen, I. Vuorinne, P. Pellikka, T. Vesala, and A. Kübert (2025): Ecosystem-scale crassulacean acid metabolism (CAM) gas exchange of a sisal (Agave sisalana) plantation, Agriculture, Ecosystems & Environment 381, 109435, doi: 10.1016/j.agee.2024.109435 . Spielmann, F.M., G. Wohlfahrt, A. Hammerle, F. Kitz, M. Migliavacca, G. Alberti, A. Ibrom, T. S. El-Madany, K. Gerdel, G. Moreno, O. Kolle, T. Karl, A. Peressotti, and G. Delle Vedove (2019): Gross primary productivity of four European ecosystems constrained by joint CO2 and COS flux measurements, Geophysical Research Letters 46 .10, 5284–5293. Spielmann, F. M, A. Hammerle, F. Kitz, K. Gerdel, and G. Wohlfahrt (2020): Seasonal dynamics of the COS and CO 2 exchange of a managed temperate grassland, Biogeosciences 17 .16, 4281–4295. Stimler, K., S. A. Montzka, J. A. Berry, Y. Rudich, and D. Yakir (2010): Relationships between carbonyl sulfide (COS) and CO 2 during leaf gas exchange, New Phytologist 186 .4, 869–878. Stoy, P. C., T. S. El-Madany, J. B. Fisher, P. Gentine, T. Gerken, S. P. Good, A. Klosterhalfen, S. Liu, D. G. Miralles, O. Perez-Priego, A. J. Rigden, T. H. Skaggs, G. Wohlfahrt, R. G. Anderson, A. M. J. Coenders-Gerrits, M. Jung, W. H. Maes, I. Mammarella, M. Mauder, M. Migliavacca, J. A. Nelson, R. Poyatos, M. Reichstein, R. L. Scott, and S. Wolf (2019): Reviews and syntheses: Turning the challenges of partitioning ecosystem evaporation and transpiration into opportunities, Biogeosciences 16.19, 3747–3775. Sun, W., K. Maseyk, C. Lett, and U. Seibt (2018): Stomatal control of leaf fluxes of carbonyl sulfide and CO 2 in a Typha freshwater marsh, Biogeosciences 15 .11, 3277– 3291. Sun, W., K. Maseyk, C. Lett, and U. Seibt (2024): Restricted internal diffusion weakens transpiration–photosynthesis coupling during heatwaves: Evidence from leaf carbonyl sulphide exchange, Plant, Cell & Environment 47 .5, 1813–1833. Vesala, T., T. Suni, Ü. Rannik, P. Keronen, T. Markkanen, S. Sevanto, T. Grönholm, S. Smolander, M. Kulmala, H. Ilvesniemi, R. Ojansuu, A. Uotila, J. Levula, A. Mäkelä, J. Pumpanen, P. Kolari, L. Kulmala, N. Altimir, F. Berninger, E. Nikinmaa and P. Hari. (2005). Effect of thinning on surface fluxes in a boreal forest. Global Biogeochemical Cycles , 19 (2), https://doi.org/10.1029/2004GB002316. Vesala, T., K-M. Kohonen, L. M. J. Kooijmans, A.P. Praplan, L. Foltýnová, P. Kolari, M. Kulmala, J. Bäck, D. Nelson, D. Yakir, M. Zahniser, and I. Mammarella (2022): Long-term fluxes of carbonyl sulfide and their seasonality and interannual variability in a boreal forest, Atmospheric Chemistry and Physics 22 .4, 2569–2584. Vuorinne, I, J. Heiskanen, M. Maghenda, L. Mwangala, P. Muukkonen & P. K. E. Pellikka (2021a): Allometric models for estimating leaf biomass of sisal in a semi-arid environment in Kenya. Biomass and Bioenergy 155, 106294. Vuorinne, I., J. Heiskanen & P. Pellikka (2021b): Assessing leaf biomass of Agave sisalana using Sentinel-2 vegetation indices. Remote Sensing , 13, 233. Wachiye, S., L. Merbold, T. Vesala, J. Rinne, S. Leitner, M. Räsänen, I. Vuorinne, J. Heiskanen, and P. Pellikka (2021): Soil greenhouse gas emissions from a sisal chronosequence in Kenya, Agricultural and Forest Meteorology 307 Crossref Google Scholar 4. Wehr, R., R. Commane, J. W. Munger, J. B. McManus, D. D. Nelson, M. S. Zahniser, S. R. Saleska, and S. C. Wofsy (2017): Dynamics of canopy stomatal conductance, transpiration, and evaporation in a temperate deciduous forest, validated by carbonyl sulfide uptake, Biogeosciences 14 , 389–401. Wehr, R. and S. R. Saleska (2021): Calculating canopy stomatal conductance from eddy covariance measurements, in light of the energy budget closure problem, Biogeosciences 18 .1, 13–24. Winter, K. and J. A. C. Smith (1996): “An introduction to crassulacean acid metabolism. Biochemical principles and ecological diversity”, Crassulacean acid metabolism: biochemistry, ecophysiology and evolution , Springer, pp. 1–13. Winter, K. (2019): Ecophysiology of constitutive and facultative CAM photosynthesis, Journal of experimental botany 70 .22, 6495–6508. jxb/erz002. Wohlfahrt, G., F. Brilli, L. Hörtnagl, X. Xu, H. Bingemer, A. Hansel, and F. Loreto (2012): Carbonyl sulfide (COS) as a tracer for canopy photosynthesis, transpiration and stomatal conductance: potential and limitations, Plant, cell & environment 35 .4, 657–667. Wohlfahrt, G., K. Gerdel, M. Migliavacca, E. Rotenberg, F. Tatarinov, J. Müller, A. Hammerle, T. Julitta, F. M. Spielmann, and D. Yakir (2018): Sun-induced fluorescence and gross primary productivity during a heat wave, Scientific reports 8 .1, 1–9. Yang, F., R. Qubaja, F. Tatarinov, E. Rotenberg, and D. Yakir (2018): Assessing canopy performance using carbonyl sulfide measurements, Global Change Biology 24 .8, 3486– 3498. Yang, X., J. C. Cushman, A. M. Borland, E. J. Edwards, S. D. Wullschleger, G. A. Tuskan, N. A. Owen, H. Griffiths, J. A. C. Smith, H. C. De Paoli, D. J. Weston, R. Cottingham, J. Hartwell, S. C. Davis, K. Silvera, R. Ming, K. Schlauch, P. Abraham, J. Ryan Stewart, H-B. Guo, R. Albion, J. Ha, S. D. Lim, B. W. M. Wone, W. C. Yim, T. Garcia, J. A. Mayer, J. Petereit, S. S. Nair, E. Casey, R. L. Hettich, J. Ceusters, P. Ranjan, K. J. Palla, H. Yin, C. Reyes-García, J. L. Andrade, L. Freschi, J. D. Beltrán, L. V. Dever, S. F. Boxall, J. Waller, J. Davies, P. Bupphada, N. Kadu, K. Winter, R. F. Sage, C. N. Aguilar, J. Schmutz, J. Jenkins and J. A. M. Holtum (2015): A roadmap for research on crassulacean acid metabolism (CAM) to enhance sustainable food and bioenergy production in a hotter, drier world, New Phytologist 207 .3, 491–504. Zahn, E., E. Bou-Zeid, S. P. Good, G. G. Katul, C. K. Thomas, K. Ghannam, J. A. Smith, M. Chamecki, N. L. Dias, J. D. Fuentes, J. G. Alfieri, H. Kwon, K. K. Caylor, Z. Gao, K. Soderberg, N. E. Bambach, L. E. Hipps, J. H. Prueger, and W. P. Kustas (2022): Direct partitioning of eddy-covariance water and carbon dioxide fluxes into ground and plant components, Agricultural and Forest Meteorology 315 , 108790. Crossref Google Scholar Information & Authors Information Version history V1 Version 1 13 May 2025 Peer review timeline Published Agricultural and Forest Meteorology Version of Record 1 Mar 2026 Published Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords cam eddy covariance stomata stomatal conductance transpiration Authors Affiliations Kukka-Maria Kohonen 0000-0001-9258-1225 [email protected] ETH Zurich Institute for Agricultural Sciences Department of Environmental Systems Science View all articles by this author Angelika Kübert Helsingin yliopiston Institute for Atmospheric and Earth System Research View all articles by this author Lutz Merbold Agroscope Forschungsbereich Agrarokologie und Umwelt View all articles by this author Matti Räsänen 0000-0003-0994-5353 Helsingin yliopiston Institute for Atmospheric and Earth System Research View all articles by this author Nina Buchmann 0000-0003-0826-2980 ETH Zurich Institute for Agricultural Sciences Department of Environmental Systems Science View all articles by this author Ivan Mammarella 0000-0002-8516-3356 Helsingin yliopiston Institute for Atmospheric and Earth System Research View all articles by this author Petri Pellikka Helsingin yliopisto View all articles by this author Timo Vesala 0000-0002-4852-7464 Helsingin yliopiston Institute for Atmospheric and Earth System Research View all articles by this author Metrics & Citations Metrics Article Usage 321 views 207 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Kukka-Maria Kohonen, Angelika Kübert, Lutz Merbold, et al. Tracking canopy conductance and transpiration of CAM-plants Agave sisalana with carbonyl sulfide fluxes. Authorea . 13 May 2025. DOI: https://doi.org/10.22541/au.174711610.06004106/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.174711610.06004106/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a004e68f8c010700',t:'MTc3OTU0ODExNg=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();