Main

Soils are highly complex environments harbouring an enormous diversity of organisms. Besides fungi, bacteria and protists are the most abundant and diverse soil organisms (Fierer and Jackson, 2006; Geisen et al., 2015) that interact in many different ways. Protists are key predators of bacteria (de Ruiter et al., 1995; Bonkowski, 2004) and they shape bacterial communities through selective feeding (Bonkowski and Brandt, 2002; Rosenberg et al., 2009). Protists recognize prey quality when they are in direct contact through bacterial morphological differences and soluble compounds (Jousset, 2012). However, it remains unknown whether protists sense their prey over long distances in the porous soil environment.

Due to their physico-chemical properties including low molecular weight, lipophilicity, high vapour pressure and low boiling points, volatile organic compounds (VOCs) can diffuse through air- and water-filled pores (Effmert et al., 2012). Consequently, they play a key role in interactions between physically separated soil microbes (Kai et al., 2009; Garbeva et al., 2014; Schmidt et al., 2015; Schulz-Bohm et al., 2015). Most bacteria produce a broad spectrum of VOCs, which are of fundamental ecological importance in cross-kingdom interactions with, for example, plants, fungi and nematodes (Gu et al., 2007; Kai et al., 2009; Effmert et al., 2012). However, whether protists as the most important predators of bacteria can sense bacterial VOCs and whether this information translates to prey suitability in specific bacteria−protist interactions remain largely unknown.

The responses of three different soil protists to VOCs emitted by six phylogenetically distinct soil bacteria were tested using a two-compartment Petri-dish system (Supplementary Material 1,Supplementary Figure S1, Supplementary Tables S1 and S2). Our results revealed that bacterial VOCs significantly altered protist activity (Figure 1), that is, higher relative abundance of trophozoites compared to cysts (Supplementary Figure S1), as well as motility and growth (Supplementary Figures S2 and S3), demonstrating that VOCs are key components in long-distance communication between protist predators and bacterial prey. To evaluate whether VOC-mediated responses reflect the outcomes of direct trophic interactions, we compared the impact of volatiles on protist activity with their responses in direct trophic interaction assays (Supplementary Material 1). In most cases, volatile-induced increases of protist activity were mirrored by an increase in activity in direct trophic interactions and vice versa (Figure 1). For example, the activity of Vermamoeba and Saccamoeba was reduced by Dyella, whereas Collimonas stimulated the activity of Vermamoeba and Tetramitus in VOC-mediated as well as in direct trophic interactions. This suggests that protists can sense suitable prey, based on bacterial species-specific VOCs. Specific long-distance diffusing bacterial VOCs can therefore provide early information about suitable prey and, consequently, more efficient predation.

Figure 1
figure 1

Effect (mean±s.e.) of six bacterial strains on the relative abundance of active protists (Supplementary Figure S1) in three species (a–c) in volatile-mediated (red bars) and direct trophic (blue bars) interactions. Relative abundances were standardized (using Z-scores) per protist species and interaction type (direct vs VOCs) and centred on the control treatment for each species. Positive values indicate more active protists relative to cysts than in the control (that is, no bacterial volatiles or no bacteria in trophic interaction assays), negative values vice versa. Different letters indicate significant differences among bars tested per type of interaction (that is, letters a−c for volatile-mediated interactions, and q−t for direct trophic interactions). Asterisks indicate significant differences from the control, while dashed bars indicate a significant difference between volatile-mediated and direct trophic interactions. For raw data see Supplementary Figures S2 and S3; for statistical analyses Supplementary Tables S3-S5.

Full size image

However, opposing patterns were also observed. For instance, the presence of VOCs from Burkholderia and Paenibacillus reduced the activity of Vermamoeba and Saccamoeba, whereas these protists increased in activity and abundance when directly preying on Burkholderia and Paenibacillus (Figure 1 and Supplementary Figure S3). Similarly, VOCs of Pseudomonas stimulated the activity of Tetramitus, whereas the bacterium inhibited this protist species in direct trophic interactions (Figure 1 and Supplementary Figure S3). Opposing effects observed in VOC-mediated and direct trophic interactions support the idea that bacteria and protist predators engage in a complex chemical warfare (Mazzola et al., 2009; Jousset, 2012). Thereby, besides producing soluble toxic compounds in the presence of the predator (Mazzola et al., 2009, Pedersen et al., 2011; Jousset, 2012), bacteria can repel potential predators by volatiles, which is in line with a report revealing inhibition of the protist Acanthamoeba castellanii by bacterial VOCs (Kai et al., 2009).

Similarly to bacteria−bacteria and bacterial−fungal VOC-mediated interactions (Schmidt et al., 2015), the response of protists to bacterial volatiles was strongly dependent on the interacting partners (Figure 1, Supplementary Figures S2 and S3). VOCs of Burkholderia, Dyella and Paenibacillus inhibited Vermamoeba and Saccamoeba, whereas they stimulated the activity of Tetramitus. The bacteria tested in this study produce distinct blends of volatiles (Garbeva et al., 2014; Schulz-Bohm et al., 2015) that can explain the varying responses of the protist taxa. Species-specific bacteria−protist interactions are in line with differential feeding (Glücksman et al., 2010) and the sensitivity to soluble (toxic) bacterial secondary metabolites of protist taxa (Pedersen et al., 2011).

Bacterial volatiles belong to different chemical classes, including alkenes, ketones, sulphides and terpenes (Lemfack et al., 2014; Schmidt et al., 2015), the latter group being especially large and diverse. Recently, Song et al. (2015) showed that the Collimonas strain used in this study emits volatile monoterpenes (β-linalool and β-pinene) and sesquiterpenes (germacrene d-4-ol and δ-cadinene). To disentangle the contribution of specific groups of bacterial volatiles such as terpenes in bacteria−protist interactions, we created Collimonas strains mutated in the terpene synthase gene and phytoene synthase gene (Supplementary Material 1) and exposed protists to the VOCs of those mutants. For the terpene synthase mutant, the activities of Vermamoeba and Tetramitus were similar to the control level and significantly reduced compared to the wild type, demonstrating a loss of function in the mutant (Figure 2). Similar effects were observed for the motility of Vermamoeba and Tetramitus (Supplementary Figure S4). VOCs of the phytoene synthase mutant did not significantly change the activity of Vermamoeba and Tetramitus compared to the wild type (Figure 2). These results suggest that terpenes play a key role in VOC-mediated bacteria−protist interactions as catalysers of protist activity, which is in line with terpene-induced stimulations of nematodes (Rasmann et al., 2005). To date the ecological and the biological function of bacterial terpenes remains largely unknown. Most terpenes volatilize easily and, thus, can travel fast and over long distances through both, the liquid and gaseous phase of the soil (Hiltpold and Turlings, 2008). Hence, terpenes may be of great ecological importance for the interactions between spatially distant soil organisms.

Figure 2
figure 2

Effect (mean±s.e.) of volatiles of wild-type Collimonas pratensis Ter91 and two mutants (Δ2617: no terpene synthase activity; Δ3184: no phytoene synthase activity) on the activity of three protist species (a−c). Phytoene synthase mutant (Col. Δ3184) and terpene synthase mutant (Col. Δ2617) prevent production of non-volatile and volatile terpenes, respectively. Protist activity was standardized (using Z-scores) per protist species and centred on the control treatment for each species. Positive values indicate higher protist activity than in the control (that is, no bacterial volatiles), negative values vice versa. Different letters indicate significant differences among bars. Asterisks indicate significant differences from the control. For raw data see Supplementary Figure S4, for statistical analyses Supplementary Table S6.

Full size image

In conclusion, we show that volatiles are key drivers of species-specific bacteria−protist interactions and that terpenes are among the informative compounds that enable protists to sense suitable prey bacteria. Species-specific responses of protists to bacterial VOCs suggest potential co-evolutionary dynamics in predator−prey interactions. Our study further suggests that specific volatiles can be used to activate protist predators of soil-borne disease agents, which might serve as a new method for biocontrol. Further work should aim at investigating the mechanisms and importance of volatile-mediated interactions in more complex settings, including natural conditions in soils.