Species‐specific responses to combined water stress and increasing temperatures in two bee‐pollinated congeners (Echium, Boraginaceae)

Abstract Water stress and increasing temperatures are two main constraints faced by plants in the context of climate change. These constraints affect plant physiology and morphology, including phenology, floral traits, and nectar rewards, thus altering plant–pollinator interactions. We compared the abiotic stress responses of two bee‐pollinated Boraginaceae species, Echium plantagineum, an annual, and Echium vulgare, a biennial. Plants were grown for 5 weeks during their flowering period under two watering regimes (well‐watered and water‐stressed) and three temperature regimes (21, 24, 27°C). We measured physiological traits linked to photosynthesis (chlorophyll content, stomatal conductance, and water use efficiency), and vegetative (leaf number and growth rate) and floral (e.g., flower number, phenology, floral morphology, and nectar production) traits. The physiological and morphological traits of both species were affected by the water and temperature stresses, although the effects were greater for the annual species. Both stresses negatively affected floral traits, accelerating flower phenology, decreasing flower size, and, for the annual species, decreasing nectar rewards. In both species, the number of flowers was reduced by 22%–45% under water stress, limiting the total amount of floral rewards. Under water stress and increasing temperatures, which mimic the effects of climate change, floral traits and resources of bee‐pollinated species are affected and can lead to disruptions of pollination and reproductive success.

The reduced water uptake associated with water stress disrupts plant metabolism. Photosynthesis and physiological processes are affected by water stress, which also reduces leaf number and stomatal conductance, and/or induces stomatal closure (Mittler, 2006). However, metabolic rates increase with increasing temperatures, up to the optimal temperature for a given plant species. Higher temperatures enhance photosynthesis by increasing stomatal conductance (Zandalinas et al., 2018). However, when increasing temperatures and water stress are combined, photosynthetic activity declines for several reasons, including decreased Rubisco activity (Awasthi et al., 2014), damage to photosystem II (Devasirvatham, Tan, & Trethowan, 2016), and increased respiration rate and high leaf temperature (Mittler, 2006). The decreased photosynthetic activity, in turn, reduces the available resources for flower development and reproduction.
Our study focused on floral biology modifications to predict the attractiveness of entomophilous plant species under abiotic stresses. We choose two bee-pollinated, Boraginaceae species: Echium plantagineum, an annual, and Echium vulgare, a biennial. Both species flowered for at least 5 weeks and produced more than 300 flowers per plant with large amounts of nectar (more than 0.3 mg of sugar per flower), allowing us to easily measure changes in floral biology. To understand the whole-plant coordinated responses, we compared the physiology, vegetative and reproductive morphology, and nectar reward production of these two species when grown under combined stress conditions (water stress and increasing temperatures). We addressed the following questions: (a) Do the changes in vegetative and reproductive morphology differ between species? (b) Do these modifications lead to a decrease in floral reward production and/or a modification of floral traits and attractiveness for bees?

| Plant material
Echium plantagineum is a late spring annual species native from the South European Mediterranean region. Echium vulgare is a biennial or a short-lived perennial native from the temperate Northern European regions. They are increasingly used in bee-friendly gardens in temperate Europe. Moreover, E. plantagineum is tested in North America as a new crop in support for pollinators in intensive agricultural landscapes (Thom et al., 2016). Echium plantagineum develops a 4-leaf rosette and a branched flowering stem in one season.
Echium vulgare produces a 20-leaf rosette during the first year of growth and one flowering stem during the second year (Klemow, Clements, & Threadgill, 2002;Piggin, 1982). Plants of both species are 20-60 cm tall. Axillary stems are produced only in the annual species. The inflorescence and flower morphology are similar. For both species, floral stem develops more than 10 scorpioid cymes which include 20-30 showy 5-merous campanulate-tubular flowers. Flowers are hermaphroditic. These two entomophilous species are mainly pollinated by bumblebees, honeybees, and solitary bees (Eberle et al., 2014;Klemow et al., 2002).
To observe the effects of temperature and water stress (and their interaction) on vegetative and reproductive development and photosynthesis-related parameters, fifteen plants per treatment and species were placed under three temperature regimes (21/19°C, 24/22°C, and 27/25°C day/night) and two watering regimes (well-watered compared to water-stressed). The well-watered plants received daily watering (soil humidity about 25%, as determined using a Procheck Hand-held Sensor 10 HS moisture sensor, Decagon Devises, Inc), whereas the water-stressed plants were watered twice a week (soil humidity of 8%-15%). The combination of temperature and watering regimes resulted in six treatments: 21°C well-watered (21WW), 21°C water-stressed (21WS), 24°C well-watered (24WW), 24°C water-stressed (24WS), 27°C well-watered (27WW), and 27°C water-stressed (27WS). In total, 90 plants per species were monitored in three growth chambers.
The photoperiod was set to 16L:8D, and relative humidity was maintained at 80 ± 10%. Growth chamber experiments lasted for 6 weeks.
Water stress was applied after 1 week of acclimation to the growth chambers; this initial week was considered week 0.

| Morphological traits
At week 0, flowering stem height was measured. Every week for 6 weeks, the number of axillary stems (for E. plantagineum), new leaves (>2 cm), inflorescences, and flowers at anthesis was counted per plant. At the end of the experiment (week 5), the height of the main flowering stem was measured to calculate the growth rate.

| Physiological traits
The 5th-node leaves of 10 plants per treatment were measured at the beginning of the experiment and 2 weeks after inducing stress.
The chlorophyll content index (CCI) was measured using a chlorophyllometer (Opti-Sciences, CCM-200), and three measurements were taken per leaf. An automatic porometer (AP4 System, Delta-T Devices) was used to measure the stomatal conductance. Gas exchange was measured using an infrared gas analyzer (IRGA ADC BioScientific LCI-SD system, serial No. 33413). The instantaneous water use efficiency (WUE i ) was calculated as WUE i = A i /E i .

| Floral and nectar traits
The corolla depth and diameter were measured three times, at weeks 1, 3, and 5, on 10 random flowers in each treatment. In week 3, flowers were dissected, and floral organs were scanned (Ricoh MP C3004 ex PS). The corolla surface area and the length of all stamens per flower were calculated using ImageJ software.

| Statistical analyses
The responses of the two species under both stresses were assessed by Principal Component Analysis (PCA). The normality of the data was estimated using QQ plots and a Shapiro-Wilk test. Physiological and morphological traits were compared between the two species under control conditions (21WW treatment) using a one-way analysis of variance (ANOVA type I). Results for all treatments were presented as relative differences compared with the control treatment 21WW for each species. The relative difference was obtained by subtracting the value of the 21WW treatment from the value of each treatment, divided by the value of the 21WW treatment. This method allowed a comparison of the responses of both species under the two stresses and their interaction.
To evaluate the effects of water and temperature stresses, linear mixed models and ANOVA type II were performed using three fixed factors (temperature × water × week) and plants as the repeated factor. Linear mixed models were used to analyze repeated measurements over time on the same plants. ANOVA type II was performed to analyze data at each time point. All analyses were performed in R 3.5.2, using the "car" package for F test, "lme4" package for linear mixed models, and "FactomineR" package for PCA. Data are presented as means ± standard errors (SE).

| Differences in physiology and morphology between the two Echium species
To obtain a global overview of the responses of the two species to water and temperature stresses, we conducted a PCA of the vegetative, physiological, and floral parameters. The first two axes of the PCA explained 52.2% of the variance (Figure 1). Axis 1 highlighted the differences between the two species and separated them based on differences in physiology (chlorophyll content and PSII efficiency), morphology (leaf number and corolla surface area), and nectar rewards (total sugar content). In the absence of stress (21WW), the annual E. plantagineum scored higher than the biennial species E. vulgare for morphological characteristics and for some physiological traits (Table 1). The annual species also produced larger flowers than the biennial species, but less nectar with a lower sugar concentration (Table 1).
Both species showed substantial responses to increasing temperatures and water stress (Figure 1a,b). Axis 2 highlighted the influence of stress on floral parameters when compared to nonstressful conditions. Flower size (diameter, depth, and surface area of the corolla) decreased under both stresses for both species. The E. plantagineum response range was broader than that of E. vulgare.

| The influence of temperature and water stresses on vegetative morphology
We compared the influence of temperature and water stresses on vegetative morphology for both species. The vegetative growth of E. plantagineum was negatively affected mainly by water stress (Table 2). The number of leaves on the main stem decreased under water stress and increasing temperatures, particularly at 27WS (Figure 2a), and the growth rate of the main stem was significantly lower for water-stressed plants (105 ± 19%) than for well-watered plants (160 ± 32%; Table 2). However, 1 week after the stress imposition, E. plantagineum plants still produced axillary stems ( Figure 2c) and initiated new leaves on these axillary stems (Figure 2d). Three weeks later, the number of leaves on axillary stems at 27WW was significantly higher than at 21WW (F 5,54 = 3.49; p = .008). Thereafter, the number of leaves decreased at 27°C, whereas it continued to increase at 21°C and remained constant at 24°C (Figure 2d). Water stress reduced the number of axillary stems and the number of leaves on those stems at all temperatures (Figure 2c,d).
The response of E. vulgare plants was different: The number of leaves on the main stem decreased significantly at increasing temperatures but was not affected by water stress (Figure 2b; Table 2).
Neither of the stresses influenced stem growth rate, which reached 41 ± 15% regardless of the treatment (5 weeks after stress imposition; Table 2). Echium vulgare maintained its growth while exhibiting foliar senescence, whereas E. plantagineum exhibited reduced growth and foliar senescence but simultaneously initiated new leaves.
The two species had different physiological responses to increasing temperatures and water stress. Chlorophyll content was significantly reduced in E. plantagineum in response to increasing temperatures, whereas it was significantly reduced in E. vulgare in response to water stress (Table 2; Figure 3c). For both species, increasing temperatures but not water stress significantly decreased the efficiency of photosystem II (Table 2; Figure 3a).
Increasing temperatures affected stomatal conductance and water use efficiency (WUE) for both species, whereas water stress significantly affected these parameters only for E. plantagineum (Table 2; Figure 3b,d). However, the effects of the stresses differed between the species. WUE increased under stress in E. plantagineum and decreased in E. vulgare compared to 21WW (Figure 3d). In the two species, both light-dependent and light-independent photosynthesis reactions were affected by stresses, and mainly by increasing temperatures.

| Floral morphology
Increasing temperature and water stress had a negative impact on flower morphology in both species. Corolla surface area decreased with increasing temperatures only in E. plantagineum, whereas it decreased in both species under water stress (Tables 3 and 4). Under combined water and temperature stress conditions (27WS), the corolla surface area for E. plantagineum decreased to about 30% of the control (21WW; 150 ± 15 vs. 515 ± 34 mm 2 ) and for E. vulgare to about 61% of the control (119 ± 12 vs. 195 ± 8 mm 2 ) ( Table 4). The mean stamen length was negatively affected by increasing temperatures and water stress in both species: stamen length decreased with increasing stress intensity in E. plantagineum, whereas it mainly decreased under water stress in E. vulgare (Tables 3 and 4). Corolla depth and diameter were also negatively impacted by both temperature and water stress, with greater reductions in E. plantagineum than in E. vulgare ( Figure 5; Table 3). The range of response was larger in E. plantagineum than in E. vulgare for floral traits: The difference in corolla surface area, depth, and diameter between the control (21WW) and the most stressful treatment (27WS) was greater for E. plantagineum than for E. vulgare.

| Nectar rewards
Temperature and water stress did not significantly decrease nectar production in E. vulgare but did in E. plantagineum (Tables 3 and 4).  (Table 4).

| D ISCUSS I ON
The annual species, E. plantagineum, was more affected by increasing temperatures and water stress compared to the biennial, E. vulgare ( Figure 6). For both species, increasing temperatures negatively affected photosynthesis parameters and both stresses reduced flower size. A major difference between the two species concerned nectar E. plantagineum, this effect was reinforced by water stress. Leaf senescence can be induced by temperature and water stresses (Sivakumar & Srividhya, 2016;Wu et al., 2010;Xu & Huang, 2007).
However, E. plantagineum compensated for this foliar senescence by initiating new leaves on axillary stems, which was not the case for E. vulgare. The production of new leaves on the main stem was particularly high in the 27WW treatment between weeks 2 and 4, consistent with previous reports that increasing temperatures can promote leaf development up to a specific optimum temperature (Gray & Brady, 2016).
We observed that increasing temperatures tended to increase the total number of flowers in E. plantagineum but did not affect flower production in E. vulgare. This result is in contrast to several studies that reported a temperature stress-induced reduction of flower production for both annual and perennial species (Liu, Mu, Niklas, Li, & Sun, 2012;Takkis et al., 2018). For both E. plantagineum and E. vulgare, water stress resulted in a decrease in the total number of flowers, and consequently in the overall floral display, with a greater reduction in the annual E. plantagineum. Similar results under water stress conditions have been reported in previous studies (Al-Ghzawi, Zaitoun, & Gosheh, 2009;Phillips et al., 2018). On the contrary, Mertensia ciliata maintained its floral display under water stress because this species is able to use stored resources to restart its spring growth; therefore, the effects of water stress are only felt after several consecutive years of drought (Gallagher & Campbell, 2017). Flowering phenology also responded differently to stress in the two species. Phenology was mostly unaffected by stress in E. vulgare, compared to the relatively large differences ob- Flower size (corolla surface area, depth, and diameter) was reduced by both stresses in our two species. Echium vulgare flowers (at 27WS) were on average two times smaller and E. plantagineum flowers five times smaller than control flowers (at 21WW). Reduced flower size (sepals, petals, and stamens) under stress has already been reported for annuals (Descamps et al., 2018;Waser & Price, 2016), biennials, and perennials (Carroll et al., 2001;Gallagher & Campbell, 2017;Halpern, Adler, & Wink, 2010;Opedal, Listemann, & Albertsen, 2016). Producing smaller flowers, which lose less water through transpiration and evaporation, can be advantageous during abiotic stress (Galen, 1999;Halpern et al., 2010).
Nectar volume for water-stressed E. plantagineum plants was on average five times lower (0.30 µl/flower) than that produced by well-watered plants (1.62 µl/flower at 21°C and 24°C). Several studies have shown that nectar volume decreased in water-stressed plants (Carroll et al., 2001;Gallagher & Campbell, 2017;Halpern et al., 2010;Waser & Price, 2016). These volume decreases were usually associated with an increase in nectar concentration in TA B L E 4 Effects of increasing temperatures and water stress on floral traits of Echium plantagineum and Echium vulgare (3 weeks after stress induction, except for number of flowers produced) Data are means ± SE (N = 10). Data followed by different letters for each parameter are significantly different (one-way ANOVA) at p < .05 among treatments.
At 21°C and 24°C, the WUE of E. plantagineum increased and photosynthetic activity was maintained, suggesting that carbohydrate production was also maintained. Even so, the total nectar sugar con-  Furthermore, if floral display is reduced, plant attractiveness is reduced. Nectar production of E. plantagineum was reduced by abiotic stresses making the species more vulnerable to pollination disruption than E. vulgare. Because nectar rewards attract pollinators, reduced quantities of nectar could decrease attractiveness, visitation rates, and pollination success. Such disruptions in plant-pollinator interactions include both morphological (corolla size and depth) and recognition (attractiveness linked to nectar production, VOCs emission) mismatches (Gérard, Vanderplanck, Wood, & Michez, 2020).

ACK N OWLED G M ENTS
We thank C. Buyens for technical assistance. Thanks to Plant Editors (K. Farquharson, J. Mach, N. Hofmann) for language improvement.
An earlier version was greatly improved by comments from two anonymous reviewers. This work is a part of a Ph.D. (C. Descamps) and two masters (S. Hugon, S. Marée) theses.

CO N FLI C T O F I NTE R E S T
None declared.