Origin of termites, sampling
Ten colonies (A–J) of Embiratermes neotenicus (Holmgren 1906) were used in this study. The colonies were collected in French Guiana in February 2019, March 2020, and February 2022 at three forest localities along the Road to Petit Saut (N5°04.250′ W52°58.770′–N5°04.650′ W53°01.360′) and one locality 13 km South-West from Kourou (N5°06.440′ W52°45.521′).
E. neotenicus colonies live in soil-made nests situated at the ground level of the forest, whose volume may reach up to multiple dozen liters. We collected large nests, having more than 50 l in volume. In such colonies, the founding queen has already been replaced by multiple NQs, and they contain, throughout the year, tens to hundreds of parthenogenetic NY4♀. Prior to dispersal flights which take place in July23,58, most mature colonies also produce up to thousands of male and female nymphs of sexual origin. Thus, the groups of parthenogenetic NY4♀ are for some time (January–June) diluted in the pool of future dispersers. However, not all large (and supposedly mature) colonies produce dispersers every year. This offered us the opportunity to collect colonies preparing for dispersal via hundreds of nymphs of both sexes as well as non-dispersing colonies containing tens to hundreds of supposedly parthenogenetic NY4♀ and only a few individual male nymphs.
Entire nests were kept intact in large plastic jars protected from direct sunlight in the Laboratory Hydreco (Petit Saut Dam, French Guiana) for a maximum of 2 days. The colonies were dissected immediately prior to the experiment, and the specimens needed for bioassays and for genetic and/or chemical analyses were collected. Fresh individuals were collected from previously undisturbed parts of the nest for the daily replacement of workers and soldiers in the bioassays.
Individual castes can be easily recognized based on their external anatomy. Different nymphal stages can be distinguished based on the head size, shape, and length of wing pads. Freshly molted NQs are non-pigmented (white) and non-physogastric, but can easily be separated from NY4 based on the short wing rudiments and larger compound eyes (Fig. 2). Sex of nymphs and non-physogastric neotenics was recognized under stereomicroscope based on the shapes of abdominal sternites, especially the enlarged sternite 6 in females.
In the three colonies used for queen inhibition bioassays (G–I), we verified the presence of RNERO in mature NQs and other available castes. Individuals were extracted in clean glass vials using 30 µl of n-hexane (GC-MS grade, Merck) per individual (or per approx. 50 eggs) for 10 min at room temperature. Extracts were transferred into clean vials and stored at −20 °C. We used two-dimensional gas chromatography coupled with mass spectrometry (GC×GC-TOFMS, Pegasus 4D Leco, St. Joseph, MI, USA), equipped with a combination of non-polar ZB-5MS (30 m, internal diameter 0.25 mm, film thickness 0.25 μm, Phenomenex, Torrance, CA, USA) and medium polarity BPX-50 (1.3 m, internal diameter 0.1 mm, film thickness 0.1 μm, Restek, Bellefonte, PA, USA) columns. The temperature program for the primary column was 50 °C (1 min) to 320 °C (20 min) at 8 °C/min; the secondary column was set 10 °C higher. We identified RNERO based on Kovats retention index (1566), elution time comparison with synthetic standard and EI-MS fragmentation pattern39. Relative RNERO abundance was compared by simultaneous visualization of all chromatograms from a given colony on an identical scale of detector intensities in Leco ChromaTOF software.
Origin of chemicals
(3S,6E)-nerolidol, (E,E)-α-farnesene, and (E,E)-farnesol were purchased from Merck (Darmstadt, Germany). RNERO was obtained from our previous synthesis and enantiomeric purity of (3S,6E)-nerolidol and RNERO were established by chiral GC39.
Statistics and reproducibility
The results presented here originate from the data on 10 different field colonies. For Experiment I, six colonies was used and each of them was treated as one replicate in a comparison of three dispersing vs. three non-dispersing colonies. For Experiments II and III, each colony was considered as an independent unit and statistically evaluated separately from other colonies, with five replicates per each treatment within each colony: replicate groups originated through random sampling of individuals belonging to appropriate castes during the dissection of the colonies. Each individual was only used in one single replicate of one single experiment.
To compare the cumulative numbers of new queens in Experiments I–III across the five replicates of different treatments, survival curves were constructed and compared using Mantel–Cox Test, with subsequent pairwise comparisons among treatments being performed using Mantel–Cox Test with Sidak-corrected 0.05 α value. Numbers of new NQs in individual replicate groups of each treatment were compared by means of non-parametric tests, i.e. Kruskal–Wallis rank-sum test and subsequent pairwise comparisons using Dunn’s test.
In electroantennographic measurements, non-dispersing colony J was used and tested NY4♀ randomly selected. Each nymph was only used for one stimulation series. Parametric analysis of variance was used (ANOVA) and the data was appropriately transformed prior to the analysis to comply with the assumptions of parametric testing.
Details on statistical evaluation are given below together with the design of individual experiments. All calculations were performed and graphs were generated in GraphPad Prism v 8.0 (GraphPad Software Inc., San Diego, CA, USA).
Bioassays were conducted in Laboratory Hydreco at 28 °C, elevated humidity and permanent darkness, dim laboratory light was used during the regular census of experimental groups.
Experiment I—readiness of NY4♀ to molt into NQs
The design of the experiment is summarized in Fig. 2a. We established six groups of 17 NY4♀, originating from six different colonies (A–F). Three of these colonies (A–C) were considered non-dispersing since they contained only a few tens of NY4 (51–168) and a low proportion of male NY4 (2–5%). The other three colonies (D–F) were apparently nearing dispersal and contained numerous male and female nymphs (stages 2–4). Most of these nymphs were of stage NY4 (from 667 to more than 1200) and the proportion of males among NY4 ranged from 41% to 42%.
NY4♀ were placed in plastic Petri dishes (6 cm in diameter) lined with a Whatman No. 1 Grade filter paper moistened with 0.2 ml distilled water. The groups were controlled every 6 h and newly differentiated NQs were scored and removed. Every 12 h, 0.1 ml water was added. The experiment was terminated after 72 h. Total cumulative counts of newly differentiated NQs from non-dispersing vs. dispersing colonies were compared across the timepoints using Mantel–Cox Test.
Experiments II and III—queen inhibition bioassays
Queen inhibition bioassays were performed with three different mature colonies. Two of them, colonies G and H, were typical non-dispersing colonies with 269 and 249 NY4♀, respectively, and a low proportion of males (1.5% and 1%). The third colony (I) contained a low number of NY4 (261), but a relatively elevated proportion of males (32%), suggesting that besides parthenogenetic NY4♀ also a non-negligible number of sexually produced females was present in the colony.
To establish the genetic origin of NY4♀ and new NQs in the three colonies, a subset of NY4♀ and all NQs that developed in the experiments were genotyped as described below.
In the initial experiment II with colony G, we tested (i) whether the presence of 10 mature NQs inhibits the formation of new NQs in groups of NY4♀, (ii) whether the amount of RNERO equivalent to headspace emission of 10 NQs mimics the effect of living NQs and inhibits the formation of new NQs, and (iii) whether 1 RNERO equivalent will also suppress NQ differentiation. The bioassay design is summarized in Fig. 4a. Groups of termites were established in new 9 cm plastic Petri dishes lined with filter paper moistened with 0.7 ml distilled water. Each dish contained 9 NY4♀, 20 workers, and 5 soldiers. To prevent mortality and ensure a fresh workforce allowing trophallactic feeding of dependent castes (soldiers, nymphs, queens), the workers and soldiers were replaced with new individuals (freshly collected from an undisturbed part of the nest) every 24 h. Every 6 h, the census of all individuals in the groups was performed and newly molted NQs were removed. Every 12 h, the chemical treatments were renewed, and 0.2 ml of water was added. The bioassay was terminated after 72 h. Termite groups were subjected to four treatments (control, 10NQ, 1QE, and 10QE), always with five replicates. Chemical treatments were applied as 10 µl of n-hexane solution on an 8 × 8 mm piece of filter paper, placed in the middle of the dish, allowing both evaporation of the substances to the headspace of the dish and direct contact of the termites with the paper. In control groups, the filter paper was treated with hexane only. In 10NQ treatment, 10 mature NQs from the same colony were added to the group together with the hexane-treated paper. In 1QE and 10QE treatments, the paper was treated with 1 or 10 equivalents, respectively, of the headspace emissions of RNERO by one mature NQ per 12 h, i.e. 260 or 2600 ng39.
The design of the subsequent experiment III with colonies H and I are summarized in Fig. 5a. In this experiment, we tested whether the 10QE treatment, which revealed to suppress the NQ differentiation in colony G, will be effective as an airborne signal, without direct access of the termites to its source. Therefore, the bioassay used three treatments, i.e. control, 10NQ and 10QE#; in the latter treatment, the treated filter paper was surrounded by a fine double mesh (2 cm outer diameter) made of copper wire, preventing the termites from antennal contact with the paper. Experiment III was run for 60 h. All other parameters were identical to experiment II.
The counts of newly differentiated NQs in all queen inhibition experiments were addressed from two perspectives and separately for each colony. First, we considered the cumulative counts of new NQs across replicates of individual treatments at each datapoint (every 6 h). From these counts, we constructed the survival curves with new NQs being treated as subjects experiencing an event and the non-differentiated NY4♀ as censored subjects. The survival curves were compared among all treatments using Mantel-Cox Test, and then pairwise comparisons of the survival curves for individual pairs of treatments were performed using Mantel–Cox Test with Sidak-corrected 0.05 α value. Second, the numbers of new NQs in individual replicate groups of each treatment were compared at the end of the experiment using the Kruskal–Wallis rank-sum test, followed by Dunn’s multiple comparisons among pairs of treatments.
To assess the genetic origin (sexual or parthenogenetic) of NY4♀ and NQs in the inhibition bioassays (colonies G–I), we genotyped a subset of NY4♀ and all NQs from the bioassays, together with workers and soldiers. The specimens were sampled into 96% ethanol and genotyped at 9 polymorphic microsatellite loci developed in our previous studies23,44. Primers are listed and methods are described in detail in Supplementary Methods and Supplementary Table 2.
We successfully genotyped 12 workers, 12 soldiers, 38 NY4♀, and 23 NQs from colony G, 13 workers, 12 soldiers, 44 NY4♀, and 10 NQs from colony H, and 14 workers, 9 soldiers, 65 NY4♀ and 7 new NQs from colony I. From allelic combinations in sterile castes, we inferred the genotypes of the putative pair of founding primaries. Subsequently, we were able to distinguish NY4♀ and NQs of sexual origin, carrying exclusive alleles of both parents at all loci at which they would be expected, from those that were missing the exclusive alleles of one of the parents (intuitively of the primary king) and were thus of parthenogenetic origin.
Central part of colony J, containing NQs, a primary king, a cohort of NY4♀, workers and soldiers, was transported to the laboratory in Prague. According to the relatively low number of NY4♀ and low male proportion among NY4♀, the colony was considered a non-dispersing colony. The nymphs were kept in the colony fragment with queens and sterile castes until the experiment to prevent their differentiation into NQs.
Antennal responses of NY4♀ were measured for RNERO, it’s opposite (S) enantiomer, and two structurally related sesquiterpenoids, (E,E)-α-farnesene and (E,E)-farnesol. Opened last flagellomere and brain were fixed between two Ag/AgCl electrodes containing Ringer’s solution and connected to a high impedance (1014 Ω) amplifier (Syntech, Buchenbach, Germany). The antennal preparation was placed into a stream of cleaned air, into which stimuli were injected from Pasteur pipettes with a 1.5 cm2 filter paper impregnated with 10 µl of the test solution. Odor injections were controlled by a foot switch-operated Syntech stimulus controller and maximal negative deflection was quantified using Syntech EAG software.
Two independent experiments were performed, one with 50 ng and the other with 500 ng of the four compounds in n-hexane. In each experiment, 15 antennae of 15 NY4♀ were stimulated with a series of the four stimuli in randomized order, preceded and followed with hexane-treated and non-treated papers as controls. New Pasteur pipettes were prepared for each series, and the data from each series was related to mean responses to hexane stimulations. The datasets were log2-transformed to reduce heteroscedasticity and comply with assumptions of variance equality for parametric testing (Brown-Forsythe test and Bartlett test for equal variances, and Shapiro-Wilk normality test) and then compared among treatments using ANOVA followed by Tukey post-hoc test.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.