Long-term experimental drought alters floral scent and pollinator visits in a Mediterranean plant community despite overall limited impacts on plant phenotype and reproduction

Pollinators are declining globally, with climate change implicated as an important driver. Climate change can induce phenological shifts and reduce floral resources for pollinators, but little is known about its effects on floral attractiveness and how this might cascade to affect pollinators, pollination functions and plant fitness. We used an in situ long-term drought experiment to investigate multiple impacts of reduced precipitation in a natural Mediterranean shrubland, a habitat where climate change is predicted to increase the frequency and intensity of droughts. Focusing on three insect-pollinated plant species that provide abundant rewards and support a diversity of pollinators (Cistus albidus, Salvia rosmarinus and Thymus vulgaris), we investigated the effects of drought on a suite of floral traits including nectar production and floral scent. We also measured the impact of reduced rainfall on pollinator visits, fruit set and germination in S. rosmarinus and C. albidus. Drought altered floral emissions of all three plant species qualitatively, and reduced nectar production in T. vulgaris only. Apis mellifera and Bombus gr. terrestris visited more flowers in control plots than drought plots, while small wild bees visited more flowers in drought plots than control plots. Pollinator species richness did not differ significantly between treatments. Fruit set and seed set in S. rosmarinus and C. albidus did not differ significantly between control and drought plots, but seeds from drought plots had slower germination for S. rosmarinus and marginally lower germination success in C. albidus. Synthesis. Overall, we found limited but consistent impacts of a moderate experimental drought on floral phenotype, plant reproduction and pollinator visits. Increased aridity under climate change is predicted to be stronger than the level assessed in the present study. Drought impacts will likely be stronger and this could profoundly affect the structure and functioning of plant-pollinator networks in Mediterranean ecosystems.MZMine: Pluskal, T., Castillo, S., Villar-Briones, A., & Orešič, M. (2010). MZmine 2: Modular framework for processing, visualizing, and analyzing mass spectrometry-based molecular profile data. BMC Bioinformatics, 11, 395. R Core Team. (2020). R: A Language and Environment for Statistical Computing. Vienna, Austria, https://www.R-project.org/. Funding provided by: AXA Research FundCrossref Funder Registry ID: http://dx.doi.org/10.13039/501100001961Award Number:1. Study site: CLIMED long-term drought experiment All field data were collected in February-June 2018. We used a subset of established plots that were part of the CLIMED (CLImate change effects on MEDiterranean biodiversity) long-term drought experiment situated at Massif de l'Étoile in Marseille, France (43° 22' N, 5° 25' E). This site has a typical woody shrub community dominated by three species: Quercus coccifera Linnaeus, 1753 (Fagaceae; anemophilous and a resource of very limited use to pollinators in the region; Ropars et al., 2020a), Salvia rosmarinus Spenn., 1835 (Lamiaceae; previously Rosmarinus officinalis; Drew et al., 2017), and Cistus albidus Linnaeus, 1753 (Cistaceae; Montès et al., 2008). Local cumulative precipitation between January and May 2018 (the flowering period surveyed) reached 291 mm, while the average precipitation between January and May for the period 2008-2018 was 205 mm (Marseille-Marignane meteorological station; www.infoclimat.fr). The site is equipped with 46 metallic control and 46 4 × 4 m rain-exclusion shelters established in October 2011, spaced by 1 to 30 m (Santonja et al., 2017). Plot locations were chosen randomly at the time of establishment of the long-term experiment, and were assigned at random to control or drought treatment (Montès et al., 2008). Gutters from rain-exclusion shelters in drought plots were designed to exclude up to 30 % and excluded on average (± SE) 12 ± 2% of precipitation between 2011 and 2018 at the centre of the plots; the intercepted water was carried away from the site with a pipe system. In control plots, the upside-down gutters intercepted a very small fraction of precipitation and rainfall reached the ground (Montès et al., 2008; Santonja et al., 2017). This water deficit attempts to mimic the mean predicted changes during the dry season in the Mediterranean area by the end of this century except in winter when rainfall is expected to increase (Giorgi & Lionello, 2008: averages for 2071-2100 relative to 1961-1990: December to February +0 to +10 %, March to May -10 to -20 %, June to August -20 to -30 %, September to November -0 to -10 %; Mariotti et al., 2015: averages for 2071-2098 relative to 1980-2005: December to February -0.1 to +0.2 mm/day, June to August -0.1 to -0.3 mm/day). The moderate but chronic experimental water deficit induced by the CLIMED experiment can alter plant physiology: carbon assimilation was reduced in C. albidus, and transpiration was reduced in C. albidus and S. rosmarinus but water use efficiency was not significantly changed in 2014 (Rodriguez-Ramirez, 2017). Between January and May 2018, permanent soil moisture probes (TDR100, Campbell Scientific Inc., Logan, Utah) measured soil moisture at 10, 20 and 40 cm in two control and two drought plots. For clarity we use the term drought to refer to the drought treatment in our study. We selected 10 control plots and 10 drought plots out of the 92 plots, based on: (i) where Thymus vulgaris Linnaeus, 1753 (Lamiaceae) was present (four plots for each treatment only) because it is an important resource for pollinators (Ropars et al., 2020a); and (ii) a high and similar percentage cover of C. albidus and S. rosmarinus. The chosen control and drought plots were homogeneously distributed throughout the site. We measured the percentage cover of each species in selected plots twice (February and June 2018). The percentage cover of S. rosmarinus, C. albidus and Q. coccifera and T. vulgaris was 21, 19, 15 and 0.5 % on average respectively in the 20 plots selected, and the community composition did not differ significantly between treatments throughout the long-term experiment. Despite such low diversity, this plant community is natural, and is representative of the site and of the type of dense, closed vegetation plant communities found in the region in areas where wildfires are ancient (> 10 years; Pimont et al., 2018). Thymus vulgaris, C. albidus and S. rosmarinus are all perennial, entomogamous shrub species; T. vulgaris is gynodioecious and obligate entomogamous (dichogamous; Arnan et al., 2014), while S. rosmarinus and C. albidus are self-compatible but with limited self-pollination (Hammer & Junghanns, 2020; Blasco & Mateu, 1995). A fourth shrub species, Ulex parviflorus Pourr., 1788, was also present but very rare (0.3 % percentage cover) with very few flowers during the study period, and other flowering species were even rarer. We did not observe any insect visit to these very rare species and hence excluded them from our study. 2. Floral traits involved in pollinator attraction 2.1. Floral scent sampling and GC-MS analysis We randomly selected up to 14 plant individuals per species in each treatment (control vs. drought) with a maximum of two (four for T. vulgaris) plants in the same plot. A few samples were lost during laboratory analysis, hence final sample sizes were 23 (control: 11; drought: 12) for S. rosmarinus, 22 (control: 11; drought: 11) for C. albidus, and 19 (control: 6 female, 6 hermaphroditic; drought: 5 female, 2 hermaphroditic) for T. vulgaris. Branches of the selected flowering plants bearing around 30-50, 2-3 or 100-400 flowers [1st-3rd quantiles] for S. rosmarinus, C. albidus and T. vulgaris respectively, were enclosed in a Nalophan bag (NA CAL, 30 cm × 30 cm, thickness 25 µm, volume ~ 2L; ETS Charles Frères, Saint-Étienne, France) connected to a pumping system maintaining a 1000 mL/min and a 200mL/min inlet and outlet air flows, respectively, provided by pumps (DC 12V, NMP850KNDC, KNF Neuberger SAS, France) powered by batteries (RS Pro 5Ah, 12V, RS Components SAS, France) and controlled by debit-metres (F65-SV1 Porter, Bronkhorst, France). Inlet air was first purified with activated charcoal (untreated, Mesh 4-8, Sigma Aldrich, USA) to limit the amount of volatiles from ambient air. Second, excess of humidity was removed using drierite (W.A. Hammond DrieriteTM Indicating Absorbents Mesh size 8, USA). Finally, ozone was filtered out through a fiberglass filter disk impregnated with sodium thiosulfate (Na2S2O3) following Pollmann et al. (2005) to limit oxidation of plant volatile organic compounds (VOCs). Air flow was first stabilized for 15 min (the time required to entirely renew the air inside the 2L-bags). VOCs were then adsorbed on a cartridge placed at the bag outlet for 10 min for S. rosmarinus and T. vulgaris, and 15 min for C. albidus. This protocol optimizes the signal-to-threshold ratio without exceeding the breakthrough volume of each VOC in the conditions of our experiment, which would distort the estimated relative composition of chemical profiles (Ormeño et al., 2007). The cartridges were made of glass tubes (Gerstel OD 6 mm for TDS2/3, RIC SAS, Lyon, France) filled with 0.120 g Carbotrap® adsorbent (matrix Carbotrap® B, 20-40 mesh, Sigma-Aldrich, France) then 0.050 g Tenax® Porous Polymer Adsorbent (matrix Tenax® GR, 20-35 mesh, Sigma-Aldrich) separated by glass wool and maintained in the tube by a fixing screen (Gerstel for TDS 2 ID 4.0 mm, RIC SAS, France) at the entrance side and glass wool at the exit side. To discriminate VOCs emitted by plants from possible environmental contamination, ambient air was sampled after every five plant samples using the same protocol. VOCs from four leaf-only plant samples per plant species were also measured to investigate which VOCs contribute most to floral scent versus leaf scent, enclosing branches of comparable size than the inflorescences of floral samples in collection bags. Sample cartridges were stored in a cooler immediately after collection, and transferred to a freezer at -20 °C as soon as possible. Prior to sampling, all cartridges had been cleaned in a Thermal Adsorbent Regenerator (RTA EcoLogicSense RG1301002, TERA Environment SARL, France) at 300 °C for 4 h. To reduce environmental variation from flowering phenology in scent emissions, each species was sampled over three days maximum around the flowering peak, during sunny weather and between 10:00 and 15:00. Throughout sampling, temperature and humidity were recorded inside and outside plant bags with data loggers (OM-EL-USB2-LCD, Omega Engineering Limited, UK). Plant parts inside bags were cut after sampling, dried in an oven at 50°C, and weighed after mass had stabilized (3-5 days, depending on plant species). Samples were analysed one to 20 days after sampling. VOCs were thermodesorbed (cool trap and flash heating -50 to 250 °C at 12°C/s for 10 min; TDS 3 Gerstel equipped with an autosampler TDS A Gerstel). They were analysed with a gas chromatograph coupled with a quadrupole low-resolution mass spectrometer in solvent vent and CIS splitless mode (GC 6890N; MS 5973N; Agilent) equipped with a HP-5MS non-polar capillary column (5 % phenyl-methylsiloxane; length 30 m; internal diameter 0.25 mm; film thickness 0.25 μm; Agilent 19091S-433). The temperature gradient applied to the column was 40°C for 5 min, then up to 245 °C at 3 °C/min and maintained for 2 min (total run time 75.33 min). The carrier gas was helium at 7.1 psi and 1 mL/min. Mass spectra were recorded in the scan EMV mode (EM voltage 1295 eV and scanned from m/z 40 to 400, with one scan every 0.004 min. Chromatograms were analysed with MZmine2 (version 2.18.1 developed for gas chromatography; Pluskal et al., 2010) in a 11-steps batch. Briefly, the baseline of each chromatogram was adjusted to 0, then values of m/z in each scan extracted and attributed to a peak. Peak heights (sum of m/z), areas and retention times, as well as mass spectrum at peak maximum were then exported from MZmine2 and imported into R (R Core Team, 2020; version 3.6.3) for peak identification. Retention indexes of each peak were calculated via a linear approximation built on a C5-C20 n-alkane series injected externally (Van Den Dool & Kratz, 1963). The retention indexes of 21 external standards were also verified with this method and checked against literature (Adams, 2007). The similarity between each peak's mass spectrum and reference mass spectra was calculated using the R function 'SpectrumSimilarity' (R library 'OrgMassSpecR v0.5-3'; Stein & Scott, 1994). The reference mass spectra were that of the 21 standards, and of two libraries converted in JPS format: the Adams 2007 library (Adams, 2007), and the NIST11 (NIST, 2011) library, was used when all similarity hits from the Adams 2007 library were lower than 0.7. Only similarity to reference molecules with a retention index within ± 15 of the peak's calculated retention index was calculated. Identification was processed sequentially starting with the most common VOCs and retention indexes were adjusted locally (+/- 10) at each step based on these new identifications. Peaks with highest similarity < 0.6 were discarded. Peaks with the same identity were then aligned for each species. True absence of a VOC was verified, and areas not integrated with the first MZmine2 round were manually added, similarity calculated as above, with peak discarded if similarity < 0.6 and using a smaller retention index tolerance of ± 5 (0.2 min on RT). Only VOCs previously reported as known plant volatiles in Pherobase (El-Sayed, 2019) or in a comprehensive review (Knudsen et al., 2006). Most of them had also previously been reported in oil extracts or as plant volatiles from the three study species (Katerinopoulos et al., 2005; Ormeño et al., 2007; Satyal et al., 2016). Following Campbell et al. (2019), we removed ambient air contaminants, by selecting only VOCs whose areas exceeded three times that of ambient air samples, by performing two-sample t-tests on , where A is the area of a peak (MZmine2 integration), Q and q are the inlet and outlet flows, respectively, and t is the sampling time, and using the type of sample (air vs. flower sample) as a factor, and with a false-discovery rate of 5 % to control for multiple comparisons. We also removed VOCs quantified in fewer than three samples of each species, because they added too much variance. We then removed the contribution of air samples to the retained VOCs in plant samples (each plant sample matched with the air sample taken closest in time) and calculated emission rates (in µg.h-1.gDM-1): (Sabillón & Cremades, 2001), with k the response coefficient calculated for each chemical family based on external calibration of pure standards (see below), and mDM the total dry mass of the inflorescence branch inside the sampling bag (flowers and leaves). Emission rates were then standardized by temperature (measured inside the bag during air flow stabilization and sampling; Ormeño et al., 2007; Sabillón & Cremades, 2001): where T is the temperature inside the sampling bag (in °K). 2.2. MZMine2 batch parameters 1. Raw data import Select all raw chromatogram files associated with one species 2. Filter scans Filter selected "Round resampling" "Sum duplicate intensities" = True "Remove zero intensity m/z peaks" = True 3. Crop filter Retention time: "min" = 4.5 min (remove solvents and water eluted in the first min) "max" = 64.0 min (after Docosane ~ 311 da; upper limit for molecule volatility at ambient temperature is ~ 300 da). m/z: "min" = 40 (lower limit of mass spectrometer acquisition) "max" = 315 (mass of Docosane molecular peak) 4. Baseline correction Correction method: "RollingBall baseline corrector" "wm (number of scans)" = 200 "ws (number of scans) = 5 5. Mass detection mass detector selected: "Centroid" "Noise level" = 2000 for 1st round, 1000 for 2nd round 6. Chromatogram builder "min time span (min)" = 0.05 "min height" = (same as 5.) "m/z tolerance" = 0.5 (absolute) / 0.001 (ppm) 7. Smoothing "Filter width" = 15 8. Deconvolution Algorithm selected : "Local minimum search" "Chromatographic threshold" = 0.50 "Search minimum in RT range (min)" = 0.04 "Minimum relative height" = 0.001 "Minimum absolute height" = 2000 "Min ratio of peak top/edge" = 1.2 "Peak duration range (min)": "min" = 0.05; "max" = 2.0 9. Peak merging "m/z tolerance" = 500 (absolute) / 5.0 (ppm) "RT tolerance window (number of scans)" = 15 "Use original raw data file" = False "Use detected peaks only" = False "Cumulative computing mode (TIC)" = True 10. Join Aligned (GC module) Our aim here was to align as few peaks as possible, so as to have one line per peak in the final matrix "m/z tolerance" = 0.5 (absolute) / 5.0 (ppm) "Weight for m/z" = 0.8 "Retention time tolerance" = 0.001 "Weight for RT" = 0.2 "Minimum score" = 0.7 "Use RT recalibration" = False "Use detected m/z only" = False "RT tolerance post-recalibration" = 0.4 "Export dendrogram as TXT" = False 11. Export CSV Export Peak RT, RTstart, RTend, Height, Area 2.3. External calibrations and calculation of response coefficients k Three different mixtures of pure standard chemicals (Sigma-Aldrich) were made using cyclohexane as solvent. Cartridges were loaded with 1 µL of the dilutions, except for Mixture 3 in which they were loaded with 1 µL of the solid mixture and 1 µL of the E-Caryophyllene solution. Three samples of each dilution were analysed for each mixture, in the same conditions as plant samples (see Main text). Two linear regressions were calculated: at low injected masses, and high injected masses (DIL6 and DIL5B for Mixtures 1 and 2, and DIL4 and DIL3 for Mixture 3). This is because for high injected masses the column is overloaded and this leads to a smaller than expected peak area. An intercept was used in linear regressions, because areas at low injected masses were indistinguishable no matter the dilution, showing that the quantification threshold (= intercept) was higher than lowest dilutions tested. In the conditions of the analysis it was impossible to properly quantify exact mass of VOCs below that threshold. 2.4. Nectar production For each plant individual on which we sampled floral scent we also selected five flowers (only 2.5 on average for C. albidus) to measure nectar production (as well as flower size and colour, see section 2.5.). We measured nectar standing crop (otherwise referred to as nectar production for simplicity) in each flower using 0.5 µL microcapillary tubes (Hirschmann Laborgeraete, Germany), placed in contact with nectaries. Flowers chosen were about to open for S. rosmarinus (any prior pollinator visit would have been prevented by the closed corolla) and sampled before 9:00 for C. albidus and T. vulgaris, to reduce variability due to flower ageing (3-4 days for S. rosmarinus and T. vulgaris and 1-2 days for C. albidus; authors pers. obs. and Flo et al., 2018) and nectar removal by pollinators (Flacher et al., 2020). Nectar volume was calculated (in μL/flower) by measuring the amount of liquid in a microcapillary tube with a digital calliper (Digit-Cal MK IV, Brown & Sharpe, USA). Sugar concentration was measured immediately afterwards using a hand-held refractometer (Eclipse 0-30°Brix, Bellingham & Stanley Ltd, UK) calibrated by the manufacturer. The whole volume of nectar collected was deposited on the measuring surface of the refractometer and diluted with 1 µL of pure water (measured with a 1 µL microcapillary tube) to increase the total volume and measure reproducibility. The sugar concentration (in µg/µL) and total sugar content (in µg) were calculated a posteriori using the conversion table from ° Brix to g/L by Kearns and Inouye (2003) and by applying the dilution factor and a temperature correction after the manufacturer's correction table (Flacher et al., 2020). 2.5. Measure of flower colour and size, and statistical analysis Flower colour. On the same flowers as used for nectar production, we recorded reflectance spectra from 300 to 700 nm at 45° using an optic fibre (FCR-7UV200-2 45°; Avantes, Apeldoorn, the Netherlands) connected to a Jaz portable spectrometer equipped with a Pulsed xenon light source (Ocean Optics, USA). Reflectance spectra were calibrated with a WS-2 white reflecting standard (Avantes, The Netherlands). Measurements were taken as such: for S. rosmarinus, in the middle of the lower petal; for C. albidus, in the middle of both the magenta and the yellow parts of two petals; and for T. vulgaris, on all the petals simultaneously, due to their very small size. Each reflectance spectrum was corrected by an offset calculated as the mean reflectance between 300 and 370 nm (the three species do not reflect in UV wavelengths) to correct for the bias due to the unavoidable light exposition variation during measurement. We calculated the position of each flower in the Apis mellifera hexagon colour space (Chittka, 1992), using the spectral sensitivity curves of its three photoreceptors (ultraviolet, blue and green; normalized each by their area under the curve) and the standard D65 as the illuminant spectrum: the position in the hexagon was calculated as the Euclidian distance between the stimulus and the background colour at the centre of the hexagon. For the background, we used the spectral reflectance of leaves of each species, which are the dominant flower background in the study site: we used the mean reflectance over five spectra for each species and each treatment, disregarding the natural variation in background colour. The hexagon representation illustrates how colours appear to a given visual system (Chittka, 1992). Flower size. On each flower, we mea