Say it with flowers

Abstract

The timing of reproduction is a critical determinant of fitness, especially in organisms inhabiting seasonal environments. Increasing evidence suggests that inter-plant communication plays important roles in plant functioning. Here, we tested the hypothesis that flowering coordination can involve communication between neighboring plants. We show that soil leachates from Brassica rapa plants growing under long-day conditions accelerated flowering and decreased allocation to vegetative organs in target plants growing under non-inductive short-day conditions. The results suggest that besides endogenous signaling and external abiotic cues, flowering timing may involve inter-plant communication, mediated by root exudates. The study of flowering communication is expected to illuminate neglected aspects of plant reproductive interactions and to provide novel opportunities for controlling the timing of plant reproduction in agricultural settings.

Introduction

Increasing evidence suggests that inter-plant communication plays important roles in plant survival, functioning and performance.^1,2 The most studied example is the “talking plants” phenomenon, whereby in response to herbivory, some plants not only increase their local and systemic resistance, e.g,³ but also release various metabolites, such as methyl jasmonate and green leaf volatiles, which induce the production of defensive metabolites and defense priming in their undamaged neighbors (for a review see ref⁴). While most studies on plant communication are related to aboveground signaling and cuing, recent findings demonstrate that belowground communication can enable plants to avoid wasteful allocation to competition with other organs on the same plant, e.g.^5–7 or kin (e.g,⁸ but see⁹), and to elicit preemptive responses against imminent herbivory (e.g.,^10,11) and drought.^12,13 While great attention has been given to plant communication and eavesdropping in the contexts of competition, herboviry and various stresses, no studies tested whether plant communication plays a role in reproduction.

The timing of reproduction is a fundamental determinant of fitness, especially in organisms living in seasonal environments.^14-17 In plants, flowering phenology is subjected to myriad selective factors^18,19 and is partially based on genetic specialization. (e.g.^20,21). In addition, flowering timing is strongly dependent on plastic responsiveness to the plant’s dynamic internal state and ecological settings. (e.g.,^22,23) For example, due to its vast resource requirements, allocation to reproduction might come at the expense of other functions, (e.g.,²⁴) and it is typically delayed until the plant has grown to a minimal size, (e.g.,²⁵) or has accumulated a minimal quantity of resources. (e.g.,²⁶)

In predominantly outcrossing plants, reproductive success is tightly dependent on the flowering phenology of other plants.^27,19 Asynchronous flowering can be selected for due to its positive effects on pollen and seed dispersal.^18,28 In contrast, flowering synchrony may improve visual and chemical pollinator attraction, increase the number of potential reproductive partners, and decrease inbreeding, (e.g.,²⁹) and seed predation.³⁰

Flowering timing involves numerous signaling and transduction pathways,^31–35 which are beyond the scope of the present study. Among the myriad external factors involved in flowering timing and synchrony are light levels,³⁶ temperature³⁷ and accumulated degree periods.¹⁸ In certain tropical plants, flowering synchrony is achieved by harmonized responsiveness to rainfall³⁸ or high air humidity,³⁹ and in some wind-pollinated plants, pollen is only released under low air humidity, ensuring effective pollen dispersal.⁴⁰ Perhaps the most-studied flowering stimulus is photoperiod,⁴¹ that in conjunction with endogenous circadian rhythms and additional external cues, is widely utilized by plants to induce reproduction at the most appropriate seasons.^42,43 Flowering can be also induced by exogenous biotic triggering. In some plants, low red:far-red ratios, correlated with imminent light competition, may accelerate (e.g.⁴⁴) or delay⁴⁵ flowering, and in Viola tricolor, pollen-tube growth and competition were found to be affected by the presence of neighbors.⁴⁶ Interestingly, we could find only one reported case of animal-induced flowering. Honeydew from aphids feeding on Xanthium strumarium was demonstrated to contain salicylic acid that induced flowering in Lemna gibba plants grown under non-inductive conditions.⁴⁷

Larger and denser flowering stands can increase both pollinator visitation rates and pollination efficiency^48,49 and these effects were found to be more pronounced at the patch level than at the population level.^50–52 (but see⁵³) The dependency of pollination efficiency on the flowering behavior of their immediate neighbors raises the possibility that besides their responsiveness to various abiotic cues, plant's flowering phenology might be also affected by signals and cues emitted from their neighbors. Specifically, it can be expected that where greater flowering intensity and coordination increase pollination efficiency, communicative flowering coordination would be favored by natural selection.

Here, we made a first step in testing the hypothesis that inter-plant communication affects flowering timing. Based on our previous findings on root communication (e.g.,^12,13), we investigated the possibility that inter-root communication can facilitate reproductive coordination. Using a simplified model system, we investigated the growth and flowering timing of target Brassica rapa plants grown under short-day (SD) conditions, while treated with soil leachates from source plants grown under either flowering-inductive long-day (LD) or non-inductive (SD) conditions. We predicted that in the presence of leachates from LD plants, SD plants would accelerate their flowering and demonstrate growth and architecture typical to LD conditions.

Materials and Methods

Plant material

We used mizuna, Brassica rapa L. var Nipposinica (Brassicaceae), a rosette-forming, self-compatible and mostly insect-pollinated biennial.⁵⁴ As a predominantly outcrosser, the fitness of B. rapa is assumed to strongly depend on foreign pollination and pollinator attraction, both of which could potentially benefit from tight flowering synchronization. B. rapa is a facultative long-day plant, i.e., although LD conditions accelerate its flowering, it can also flower under SD conditions, thus could potentially be triggered to flower earlier even under non-inductive SD conditions, in contrast to obligate LD plants, which might be less sensitive to external cues that contradict photoperiod cues. In addition, both plant architecture and resource allocation of B. rapa are highly responsive to photoperiodic cues, with greater allocation to vegetative growth under SD conditions and greater allocation to reproductive organs under LD conditions.⁵⁵ The genome sequencing of B. rapa has been recently completed,⁵⁶ allowing future delving into genetic and genomic aspects of the studied phenomena. Finally, the rosette-forming habit of B. rapa makes its transition to bolting readily visible and easy to monitor.

Experimental design and setup

Target plants were grown under natural SD conditions while receiving soil leachates from source plants grown under either LD or SD conditions (Fig. 1). B. rapa plants were grown from seed (Ben-Shahar seeds, Tel Aviv, Israel), in a semi-controlled greenhouse at the Sede Boqer campus, Israel (30° 50′N, 34° 46′E). Seeds were sown in 12 cm diameter pots filled with 0.8 L of commercial garden soil mixture (Deshanit, Beer Yaakov, Israel). To increase the concentration of root exudates, source plants were planted in foursomes and target plants were planted as singles, simulating a crowded stand of induced (source) plants in the vicinity of a non-induced (target) plant. Rather than simulating a realistic scenario (under field conditions, day length conditions are obviously uniform for any number of neighboring plants), this design emulates a situation where a non-induced focal plant is receiving root cues from its induced neighbors.

Figure 1. Testing for the effect of root communication on flowering acceleration. Target plants were grown under natural SD conditions while receiving soil leachates from source plants grown under either LD or SD conditions.

The Seeds of source and target plants were sown on Oct 10 and Oct 18, 2011, respectively, and differential leachate treatments started on Oct 27, when the target plants had 3–5 leaves. Leachate applications were performed every 2–3 d. Leachates were prepared by irrigating each source pot with 150 ml of tap water, 90–100 ml of which drained into a drip tray. To increase the concentration of root exudates in the leachates, source pots were kept immersed in their own 1 cm deep leachates for 60 min, after which the leachates were transferred to the assigned target pots (Fig. 1). Fifty-100 ml of leachate were slowly added to the soil surface of each target pot until reaching field capacity. Target pots were individually placed in drip trays to prevent the transfer of leachates between neighboring plants. Excess leachates were removed from the drip trays 60 min after leachate application to ensure appropriate root aeration. All plants were watered with tap water 2–3 times per week as needed, and with nutrient solution (1 gr L⁻¹ of 20:20:20 compound N, P, K plus microelements fertilizer (Ecogan, Caesarea, Israel) once a week, during the last 28 d of the experiment, in which plants have shown signs of senescence. Treatments were replicated 15 times. The experiment was terminated when all plants flowered, on March 7, 2012, 141 d after germination of the target plants, at which time all plants were harvested for further analyses. The experiment was repeated in the following year (October 2012 to March 2013), using the same experimental facility and methods, with similar results.

Lighting conditions

The experiment started 19 d after the equinox, when day length was 11:35 h long. LD and SD source plants were grown in adjacent sectors of the same greenhouse, fully separated by an opaque light barrier to avoid night-breaking light pollution in SD sector of the greenhouse (Fig. 1). Throughout, long-day conditions were imposed on LD source plants using 4 h of photosynthetically-negligible night interruption (30 µEm⁻² sec⁻¹ of cool-white fluorescent and 5 µEm⁻² sec⁻¹ of incandescent light) in the middle of the dark period.⁵⁷ Air temperature was partially controlled by an automated pad-and-fan system (Termotecnica Pericoli, Albenga, Italy), which was activated when air temperature in the greenhouse exceeded 30 °C. Daytime light availability was 45% of full sunlight.

Measurements

Plants were surveyed for bolting and flowering three times a week. Bolting and flowering times were determined by the number of days from sowing to the earliest appearance of elongation of the seminal shoot by 5 mm, and the appearance of the first open corolla, respectively. The number of leaves below the first flower on the seminal shoot was recorded at bolting in all plants. Resource storage was quantified by measuring hypocotyl size, which is an important storage organ in B. rapa and in other crucifers. (e.g.,⁵⁸) Due to its occasional asymmetry, hypocotyl size was calculated by averaging its largest diameter and the diameter perpendicular to its largest diameter. The target plants were surveyed for total number of leaves and hypocotyl diameter every 10–30 d, but these measurements could not be conducted in the source plants, the density of which did not allow accurate non-destructive measurements of these variables. At harvest, the hypocotyl diameter was recorded in the source plants and the numbers of flowering and fruiting inflorescences were recorded in all plants. Dry biomass was measured after drying the plant samples in a ventilated oven at 60 °C for at least three days.

The effects of photoperiod or leachate source on mean bolting and flowering timing, and various morphological variables were tested using one-way ANOVAs (SYSTAT 10, Chicago IL, USA). To account for their entire courses, the effects of photoperiod and leachate treatments on bolting and flowering kinetics, and the number of leaves at bolting were further analyzed using permutation tests.^59,60

Results

Source plants

'Source plants' were grown under either LD or SD photoperiod conditions and their soil leachates were used to treat target plants grown under SD conditions (Fig. 1). Day length did not affect the number of rosette or cauline leaves below the terminal inflorescence of the seminal shoot (Fig. S1), but it strongly affected reproduction timing (Fig. 2A-B). On average, LD plants bolted 13.6 d and flowered 9.4 d earlier than SD plants (P < 0.001 for both), representing 42% and 27% accelerations, within the bolting and flowering periods, in LD compared with SD plants, respectively (Fig. 2A-B). At harvest, LD plants had 222% more flowering inflorescences and 29% more fruiting inflorescences than did the SD plants (Fig. 2C). The earlier transition to reproduction in the LD plants came at the expense of vegetative growth. At harvest, hypocotyl diameter was 27.3% larger in SD than in LD plants (Fig. 2D), and rosette and hypocotyl biomasses were 81.6% and 70.3% greater in SD than in LD plants, respectively (Fig. 2E). Photoperiod did not significantly affect total plant biomass (P = 0.27), root biomass (Fig. 2E), root allocation (Fig. S2) or plant height (Fig. S3).

Figure 2. The effects of photoperiod on flowering transition and development of source plants. Data are for bolting (A) and flowering (B) times, and for inflorescence development (C), hypocotyl diameter (D) and plant biomass components (E), at harvest, of plants grown under LD (solid red lines and bars) and SD (broken black lines and white bars) conditions. Bars represent treatment means ± 1 SEM (n = 15). Curves within each graph (A-B) were compared using permutation tests.⁶⁰ Differences between treatment means (C-E) were tested using one-way ANOVAs; n.s. P > 0.05; * P < 0.05; *** P < 0.001.

Target plants

Target plants were grown under continuous SD conditions, while treated with leachates from either LD- or SD-source plants, hereafter referred to as LD- and SD-target plants, respectively.

On average, the LD-target plants bolted and flowered 9.7 and 7.5 d before the SD-target plants, respectively (P < 0.001 for both), representing 34.6% and 27.8% accelerations within the bolting and flowering periods, in LD- compared with SD-target plants, respectively (Fig. 3A-B). Plants of both treatments developed similar numbers of rosette and cauline leaves below the terminal inflorescence of the seminal shoot (Fig. S4); however, at bolting, before any leaf shedding took place, the per-plant total number of leaves was 32.8% greater in SD- than in LD-target plants (P = 0.013; Figure 4A). In the first 100 d of the experiment, plants of both treatments similarly increased their hypocotyl sizes (Fig. 4B) and total leaf numbers (Fig. 4C); however, concurringly with bolting (Fig. 3A), LD-target plants decreased hypocotyl and leaf growth, and demonstrated greater leaf senescence and shedding, compared with SD-target plants (Fig. 4C).

Figure 3. The effect of root communication on flowering transition of target plants. Bolting (A) and flowering (B) times were recorded in target plants grown under SD conditions while receiving soil leachates from LD-grown (solid red lines) or SD-grown (broken black lines) source plants. Curves within each graph (A-B) were compared using permutation tests.⁶⁰

Figure 4. The effect of root communication on morphogenesis and growth of target plants. Per-plant total number of leaves (x-axis) produced before bolting (A), kinetics of hypocotyl diameter (B) and total number of leaves (C), and inflorescence (D) and biomass components (E) were recorded at harvest in target plants grown under SD conditions while receiving soil leachates from LD-grown (solid red lines and bars) or SD-grown (broken black lines and white bars) source plants. Bars represent treatment means ± 1 SEM (n = 15). Differences between treatment means were tested using one-way ANOVAs; n.s. P > 0.05; * P < 0.05; ** P < 0.01; *** P < 0.001.

At harvest, LD-target plants had 57.8% more flowering inflorescences and 115.4% more fruiting inflorescences than did the SD-target plants (Fig. 4D). In addition, while the biomass of the cauline branches was 50% greater in LD- than in SD-targets plants, rosette and hypocotyl biomasses were 53.9% and 62.7% greater, respectively, in SD- than in LD-target plants (Fig. 4E). Leachate treatments did not significantly affect total plant biomass (P = 0.23), root biomass (Fig. 4E), root allocation (Fig. S5), or plant height (Fig. S6).

Discussion

The results suggest that flowering timing can be affected by interplant root communication. Besides an endogenous florigen, which is produced in photoperiodically-induced leaves and is triggering flowering in shoot apexes,⁴² flowering timing may also involve inter-plant cuing, whereby flowering plants exude root signals or cues that act as exogenous flowering accelerators. Although the exact identity of the cues is still unknown, the experimental design (Fig. 1) allows asserting that the observed flowering communication (FC) is mediated by root exudates. To reliably represent the reproductive phenology of the emitting plant, these compounds are speculated to be short-living and possibly rapidly decomposable by soil microorganisms, as their accumulation in the soil (e.g.⁶¹) would render them less informative and unreliable. Although the observed communication is likely to be directly mediated by root exudates, the involvement of third-party agents such as soil bacteria and non-mycorrhizal fungi cannot be ruled out. (e.g.^62,11)

Flowering coordination is expected to be beneficial under various ecological scenarios^29,30,63 and plants are known to utilize various external cues to increase reproductive synchrony.⁶⁴ However, known flowering induction cues are mostly correlated with large-scale environmental phenomena and are not necessarily representative of the specific reproductive state of immediate neighbors. Flowering coordination can be attained by common responsiveness to external cues such as day length;⁴³ however, heterogeneities in local resource availabilities, micro-climate and other growth conditions might limit the information content, relevance and resolution of such global environmental cues. Regardless of day length responsiveness, the results suggest that flowering coordination might be further tightened by inter-plant communication, which is based on the emission of and responsiveness to root exudates correlated with the specific reproductive phenology of neighboring plants. Accordingly, FC is expected to be common where synchronized flowering predictively increases fitness via various mechanisms, such as increased pollinator attraction and outcrossing probability^14,29,63 (but see¹⁹), decreased probability of seed predation^30,65 or improved seed dispersal.⁶⁶ However, the degree and prevalence of FC are hypothesized to depend on multiple evolutionary and ecological factors as follows-

Mating strategy

All other things being equal, the prevalence of FC is expected to be positively correlated with the degree of reliance of the plant’s fitness on outcrossing,^67,68 and at the extremes, to be the most prevalent among self-incompatible plants and the rarest among obligate selfers. Experiments are underway to compare the prevalence of FC in selfer-outcrosser congener pairs.

Pollinator limitation

FC is expected to be more prevalent among chronically pollinator-limited plants and where pollinator attraction is affected by local flower abundance and density.^51-53 Accordingly, communication-mediated flowering coordination is expected to be more beneficial where population structure is inherently viscous and individuals tend to be physically assorted into kin clusters, due to e.g., limited dispersal⁶⁹ or clonal aggregation. (e.g.,⁷⁰)

Pollinator competition

FC could also increase flowering asynchrony, mediated by either active flowering deterrence of plants by their early-flowering neighbors or by flowering delay of subordinate plants yielding their already-flowering neighbors. FC-mediated flowering asynchrony is expected to occur where pollination efficiency is negatively correlated with the density of the flowering plants in the patch (e.g.,^71,72) or with the time passed from the beginning of the flowering season. (e.g.,¹⁹)

Taxon identity

FC may exist between plants belonging to different taxa, but its prevalence and effects are expected to strongly depend on pollinator foraging behavior and the dependency of the plants’ fitness on pollinator fidelity,⁷³ and the extent to which pollinator attraction and pollination efficiency are affected by patch diversity, density and spacing.^74,51,52 Consequently, it is expected that interspecific FC is selected for where plants benefit from greater pollinator visitation rates in mixed plots. (e.g.,⁷⁵) Conversely, where pollination success depends on high pollinator fidelity and can be negatively affected by concurrent flowering of interspecific neighbors, (e.g.,⁷⁶) plants are expected to employ FC to increase intra-specific flowering synchrony but also actively interfere (i.e., deter, delay) with the flowering of inter-specific neighbors. Although flowering interference, which relays on ‘clogging up’ of the stigma by germinating interspecific pollen is well documented, (e.g.,^77,78) and various allelopathic effects can delay flowering, (e.g.,⁷⁹) we are not aware of existing studies demonstrating root-mediated flowering interference (“flowering allelopathy”) speculated here. Ongoing experiments are comparing the effects of FC in various intra- and inter-specific settings.

Biotic vs. abiotic pollination

Due to its potential effects on pollinator attraction, FC is expected to be more prevalent among plants with biotic than abiotic pollination. Accordingly, comparing the prevalence of FC between e.g., insect- and wind-pollinated taxa may shed light on the relative importance of pollination attraction and other selective forces for the evolution of FC.

The fact that the experimental treatments did not affect total plant biomass or root allocation supports the notion that the differential effects of the leachate treatments on the growth and development of the SD-grown target plants was not an artifactual outcome of differential nutrition or growth conditions experienced by the SD- and LD source plants, other than day length. As expected, the differential effects of the day length (source plants) and leachates (target plants) on flowering timing were accompanied by significant modifications of plant architecture and allocation to reproduction, vegetative growth and storage. These results agree with theoretical predictions made for the phenology of hypothetical annual plants with constant relative growth rates and a growing season of a fixed length. Under such conditions, plant are expected to maximize their fitness by adapting a “big-bang” ontogeny, namely, maximize the initial vegetative growth period and sharply shift to a short reproductive period at the very end of the growth season,^80,81 though decreased environmental predictability and additional factors are expected to decrease the sharpness of this transition. (e.g.,^82,83) Although the transition to reproduction is commonly accompanied by decreased vegetative growth, (e.g.^84,85) to the best of our knowledge, the current study provides the first evidence for plastic alteration of vegetative-reproductive allocation in response to interplant root cuing.

Our findings serve as a first step in describing a novel form of plant communication. Naturally, at this early stage much more work is required to discover the involved cues and testing the prevalence and relevance of FC under more realistic ecological settings. Ongoing work is concurrently focusing on testing the above-mentioned ecological hypotheses and on deciphering its underlying mechanisms. In order to identify the involved cuing vectors, we employ various metabolomic and proteomic profiling methods on root exudates and soil leachates, collected from plants grown under variable flowering-inductive and non-inductive conditions. Upon initial screening of candidate vectors, specific knockout mutants and additional techniques will be employed to pinpoint the involved signaling pathways. Further pursuing the results of the present study is expected to illuminate neglected aspects of plant reproductive ecology and to open novel possibilities to control flowering timing in agricultural settings.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Simon Barak, Pedro Aphalo, Eric von Wettberg and Bill Kunin for useful discussions and Miri Vanunu, Oron Goldstein and Laurence Bradt for technical help. The study was supported by a research grant no. 1050/11 from the Israel Science Foundation to A.N. This is publication no. 830 of the Mitrani Department of Desert Ecology.

Supplemental Material

Supplemental data for this article can be accessed on the publisher's website.