Cross-correlations of Biogenic Volatile Organic Compounds (BVOC) emissions typify different phenological stages and stressful events in a Mediterranean Sorghum plantation

Highlights

• The sorghum plantation resulted a source of BVOC, mainly methanol and acetaldehyde.

• The application of SOM analysis allowed to recognize BVOC emission signatures.

• Distinctive patterns of BVOC fluxes correlated with sorghum growth stages.

• BVOC fluxes qualitatively differentiated stressful events of lodging and cutting.

Abstract

Climate change will affect the growing season and increase the occurrence of extreme stressful events, thus altering crop phenological phases and the associated emission of biogenic volatile organic compounds (BVOC). BVOC exchange has been poorly investigated in field crops, especially in the Mediterranean area. In this study we report continuous measurements of BVOC fluxes and CO₂ net ecosystem exchange (NEE), together with environmental variables, green area index (GAI) and aboveground biomass (AGB) during a whole growing season in a grain sorghum (Sorghum bicolor x Sorghum sudangrass., cv. Nicol, Pioneer) plantation located in Southern Europe. Results of this intensive field campaign showed that, while the bare soil of our site was a sink of BVOC, the sorghum plantation became a source of oxygenated BVOC, mainly methanol and acetaldehyde, which were emitted over the season at an average rate of 0.137 ± 0.013 and 0.070 ± 0.004 nmol m⁻² s⁻¹, respectively. In addition, the application of the advanced data mining method of Self-Organizing Maps (SOM) revealed distinctive patterns of BVOC fluxes correlating with sorghum growth stages (GS): in the first stage (GS1), developing plantlets emitted a mixture of BVOC uniquely characterized by monoterpenes; in GS2, adult plants forming an homogeneous dense canopy emitted the most abundant fluxes of a mixture of oxygenated BVOC comprising methanol, acetaldehyde, formic acid, acetone, acetic acid and n-pentenol; once plants entered the flowering stage (in GS3), only a few BVOC continued to be emitted at the highest rates (i.e. formic acid, acetone, acetic acid, n-pentenol). Moreover, the application of SOM to a sub-set of BVOC fluxes highlighted the possibility to qualitatively differentiate stressful events of plant lodging and harvest cutting. In fact, enhanced emission of acetaldehyde distinguished the BVOC mixture emitted from lodged rather than from cut and harvested sorghum plants in the field.

Introduction

Climate change will impact on crops by either altering the length of the phenological phases or shifting the entire growing season (Wang et al., 2019, Ye et al., 2019, Zhang et al., 2014), thus making agricultural ecosystems more difficult to manage. Moreover, alteration of plant phenology will increasingly expose crops to extreme climatic events, especially in the Mediterranean basin where both the occurrence and intensity of stressful environmental conditions are expected to increase over the coming years (IPCC, 2019). As a consequence, the potential yields of Mediterranean crops will be negatively affected (Cramer et al., 2018) by a reduction in the amount of CO₂ assimilated trough photosynthesis and converted into biomass and grain (Dusenge et al. 2018).

Besides, plants make use of a significant part of the assimilated CO₂ to synthesize and emit blends of biogenic volatile organic compounds (BVOC) (Kesselmeier et al., 2002; Dudareva et al., 2006), which play important roles in the interaction between the different components of the ecosystem (Baldwin et al., 2006, Dicke et al., 2009). Fluxes of BVOC are regulated by physiological and physicochemical factors (Niinemets et al., 2014), and can be constitutively emitted from some plant species throughout the whole growing cycle (Bracho-Nunez et al., 2013), hence undergoing both short (i.e. daily) and long (i.e. seasonal) term variations (Oberbolz et al. 2013; Brilli et al., 2014a). In addition, abiotic and biotic stresses ubiquitously induce emissions of BVOC from plants (Kessler and Baldwin, 2001; Loreto and Schnitzler, 2010), and alterations due global change are expected to further increase the proportion of carbon allocated to the production of BVOC (Peñuelas and Staudt, 2010).

Today, the micrometeorological method of eddy covariance (Baldocchi, 2014) is employed in combination with the high throughput analytical technology of Proton Transfer Reaction ‘Time-Of-Flight’ mass spectrometry (PTR-TOF-MS) for quantifying, in quasi-real-time, a broad range of BVOC fluxes exchanged by ecosystems (Müller et al., 2010; Park et al., 2013) useful for assisting models able to estimate the impact of reactive BVOC on air quality (Guenther et al., 2012; Messina et al., 2016). Since tree species represent a substantial source of BVOC (Kesselmeier and Staudt, 1999), the vast majority of the field campaigns have been monitoring fluxes of BVOC in different types of forests worldwide, although often for limited periods of time (Karl et al., 2004; Davison et al., 2009; Langford et al., 2010; Schade et al., 2011; Schallhart et al., 2018). On the other hand, intensive measurements by eddy covariance systems equipped with multiple sensors over entire seasons generate highly resolved and highly dimensional ‘big’ data that require challenging analysis and interpretation through machine learning tools (Singh et al., 2016). Nevertheless, such advanced data mining methods have still been poorly applied to explore long-term multiple correlations between fluxes of BVOC and their dependency to both physiological and environmental variables (Luyssaert et al., 2007; Brilli et al., 2016; Savi et al., 2020). Identification of relationships between BVOC fluxes, physiological and environmental conditions could be exploited to monitor crop development in the field, as well as for diagnosing the onset of stressful events (Niederbacher et al., 2015; Kravitz et al., 2016; Brilli et al., 2019). This might help decision making on the best practises for sustainable agriculture to optimize yields (Jin et al., 2020).

So far, little attention has been given to the exchanges of BVOC by agricultural ecosystems during an entire growing season and, to our knowledge, the few reports available on crops have investigated only maize plantations (Das et al., 2003; Bachy et al., 2016), managed grasslands (Ruuskanen et al., 2011), winter wheat (Bachy et al., 2020), and a ryegrass field (Custer and Shade, 2007). Grain sorghum (Sorghum bicolor L.) is a C4 forage species that has shown to be a valid alternative crop to maize for animal feed especially in Southern European regions, as well as in areas characterized by low rainfall, especially in summer (Balole and Legwaila, 2006). In particular, previous research on maize has estimated that by maintaining the current genotypes, sowing dates, and the mix of rainfed and irrigated land use across Europe, the crop yield would result in a ~ 20% decrease by 2050 (Webber et al., 2018). Therefore, the progressive replacement of maize with drought- and heat-resistant crops has become necessary to reduce yield loss with sustainable practises.

In this study, we have measured continuously the eddy covariance fluxes of BVOC and CO₂ above a sorghum plantation located in Southern Italy during a whole growing season. In addition, we have applied the unsupervised machine learning approach of Self-Organizing Maps (SOM) to visually analyze the time-series of data in component planes where the multivariate relationships between fluxes of BVOC, net ecosystem exchange of CO₂ (NEE), indexes of crop growth (i.e. green area index, GAI; above ground biomass, AGB) and environmental variables (PAR, air temperature, VPD) can be displayed. We aim at revealing distinctive BVOC patterns featuring different phenological stages of sorghum and, at the same time, characterize the signature of BVOC emissions induced by stressful events of lodging and harvest cutting that have occurred over the season.