Cyclical drought and herbivore threats are potential causes of leaf variegation dimorphism in Cypripedium forrestii

Abstract

Leaf variegation, the mosaic of colors on the leaf surface, can be developed by certain plant species without external influence. Although it may be associated with a variety of functions, the stable existence of different leaf color morphs within a plant species has not been fully explained by previous studies. This study focuses on the two leaf morphs of Cypripedium forrestii, an endangered lady slipper orchid, and compares their micromorphological structure, photosynthetic potential, differentially expressed genes (DEGs), and ecological features to gain a comprehensive understanding of the underlying leaf variegation polymorphism. Our findings demonstrate that leaf variegation is not pathological and does not affect photosynthetic potential. Additionally, it significantly reduces herbivory damage. We found that the probability of herbivory and leaf area loss for variegated leaves was notably higher under drought conditions. Therefore, variegated individuals may be more adaptive under such conditions, while non-variegated ones may be more cost-effective in normal years. These results suggest that different leaf color morphs may be favored by varying environmental conditions, and leaf polymorphism may be a legacy of ancient climate and herbivore fluctuations.

摘要

间歇性干旱和食草动物威胁是玉龙杓兰叶斑二态性的潜在成因

某些植物在没有外部刺激的情况下会在叶面上产生不同颜色的斑纹，即植物的斑叶现象。尽管之前的研究指出斑叶现象可能具有不同的功能，但未能解释某些物种内部的叶斑多态性。本研究以濒危的玉龙杓兰(Cypripedium forrestii)为研究对象，其同时存在斑叶和无斑叶两种个体。通过比较微观形态结构、光合作用潜力、差异表达基因(DEGs)和叶片被取食情况等数据，发现玉龙杓兰的斑叶现象不是病理性的，也不影响其光合作用潜力。叶斑的存在显著减少了食草动物的损害。无斑个体被食草动物取食的概率和叶面积损失显著更高，且在干旱年份中更明显，进而使得斑叶个体在干旱环境中可能具有更大的生长优势，而无斑叶个体在正常年份可能具有更大的成本效益。上述研究结果表明，具有不同叶斑形态的个体可能受到不同环境的选择，玉龙杓兰叶斑二态性可能是长期气候波动下平衡选择的结果。

INTRODUCTION

Pathogens, abiotic stress, herbivory or mechanical damage may cause discolored patterns on leaf surfaces. However, certain plant species can also develop such a mosaic of colors without any external influence. This type of non-pathological leaf variegation, often referred to as leaf mottling, typically manifests as spots, patches, stripes and other irregular patterns of multiple colors on the leaf surface (Sheue et al. 2012; Zhang et al. 2020). Leaf variegation is commonly observed in plant species from various families in nature, with most being herbaceous understory plants or epiphytes (Givnish 1990), such as cyclamens (Konoplyova et al. 2008), Erythronium spp. (La Rocca et al. 2014), tulips of sect. Vinistriatae (Dekhkonov et al. 2022), orchids (Schuiteman 2021), begonias (Sheue et al. 2012) and certain species in the Gesneriaceae (Zhang et al. 2020). Some of these species are monomorphic, meaning all their leaves have nearly identical color patterns, while others display polymorphism, presenting two or more distinct leaf morphs within their populations.

Many research has focused on the structural mechanisms behind the phenomenon. Lev-Yadun concluded nine distinct types of visually distinct white variegation formations in leaves and other organs (Lev-Yadun 2014). Zhang et al. described five types of leaf variegation according to the different leaf structures affected: chlorophyll content, air space, epidermis, pigment and appendage (Zhang et al. 2020). Studies on non-anthocyanin-related leaf variegation have indicated that certain leaves with white-color patterns can reduce direct damage from herbivores (Cahn and Harper 1976; Campitelli et al. 2008; Wong and Srivastava 2010) or deter moth oviposition (Smith 1986; Soltau et al. 2009). The defensive role of variegation could be realized through several mechanisms, including camouflage, aposematic signaling, mimicry of insect or fungal damage, dazzle effects, reduction of plant food quality for herbivores and potentially other unidentified mechanisms (Lev-Yadun 2014, 2019; Lev-Yadun and Niemelä 2017). In addition, leaf coloration in some species may be involved in attracting pollinators, such as Columnea spp. (Gesneriaceae) (Jones and Rich 1972) and Cypripedium fargesii (Orchidaceae) (Ren et al. 2011). Nonetheless, the molecular basis of leaf variegation remains relatively understudied, and many studies are unable to fully explain the stable co-existence of leaf polymorphism within a species.

A genetic study into the dimorphic Mimulus verbenaceus found no evidence of positive selection and suggested that its leaf variegation may be non-adaptive (LaFountain et al. 2022). In contrast, Smith demonstrated that leaf polymorphism of Byttneria aculeata is correlated with light condition and herbivore stress, indicating that different leaf color morphs may be favored in different habitats (Smith 1986). However, both studies have limitations: the former lacked field monitoring and experiments, while the latter did not incorporate molecular analysis. Further investigation is required to elucidate the evolution and ecology of leaf polymorphism.

Here, we take a closer look at the genus Cypripedium. The aforementioned C. fargesii is a member of sect. Trigonopedia. This orchid lineage includes several monomorphic species characterized by heavily spotted leaves, for example, Cypripedium lichiangense and Cypripedium margaritaceum, as well as two species with non-spotted leaves, namely Cypripedium bardolphianum and Cypripedium miranthum (Cribb 1997). Species with spotted leaves, like C. lichiangense (Zheng et al. 2022), C. fargesii (Ren et al. 2011), and C. sichuanense (Li et al. 2012) produce large flowers with fungal or decomposing odors to attract larger fly species (Syrphidae and Platypezidae). In contrast, C. bardolphianum (Lu et al. 2022) and C. micranthum (Li et al. 2012) produce small, fruit-like odor flowers and are pollinated by small fruit flies (Drosophilidae).

During our field investigations, we noticed the dimorphism in the leaf variegation of Cypripedium forrestii, one of the most endangered species of sect. Trigonopedia. Two morphs of C. forrestii coexist in all populations, one with non-spotted leaves and one with blackish-purple spots on the leaves. Based on our observations of wild populations and cultivated plants, the area and color of leaf variegation remain consistent throughout the growing season. Notably, variegated leaves generally exhibit a longer lifespan, approximately 3–5 weeks longer than non-variegated leaves. Although the size and density of spots on variegated leaves may fluctuate over time, these two morphs are not interconvertible. C. forrestii is therefore a good model for the study of leaf variegation. This study concentrates on C. forrestii, and examines the micromorphological structure, photosynthesis potential, differentially expressed genes (DEGs) and ecological features underlying its leaf variegation polymorphism. By integrating these findings, we aim to address: (i) the structural and molecular differences between the two morphs; (ii) the ecological significance of leaf variegation polymorphism and (iii) the potential evolutionary process that led to the establishment of leaf polymorphism in C. forrestii.

MATERIALS AND METHODS

Plant materials

Cypripedium forrestii, endemic to northwestern Yunnan, China, has an extremely limited distribution in Lijiang (Yulong Snow Mountains) and Shangri-La city. Our previous population genetic study excluded the possibility of interspecific hybridization between C. forrestii and the sympatric species C. bardolphianum or C. lichiangense (Lin et al. 2024). The two leaf morphs found in C. forrestii are distinct phenotypes that belong to a single genetically distinct species.

As C. forrestii is a typical clonal plant, distinguishing adjacent genetic individuals in the wild is challenging. Therefore, it is crucial to clarify that our basic investigation unit is ‘one leaf’, and a pair of leaves constitutes one ‘ramet’. Each ramet can produce only a single flower. We collected samples from 28 ramets with spotted leaves and 24 ramets with non-spotted leaves of C. forrestii from eight sites in July 2022. To ensure the sampled ramets were genetically independent, we maintained a minimum distance of 10 meters between each, considering that the rhizomes of C. forrestii are typically 30–50 cm in length. The locations of these sampling sites are illustrated in Supplementary Fig. S1. The samples were categorized into three groups: non-spotted green leaves (ng), spotted purplish areas from variegated leaves (va), and non-spotted green areas from variegated leaves (vg).

Examination of micromorphological features

To investigate leaf thickness, internal leaf color, and cell arrangement in fresh C. forrestii leaves, a total of 24 small leaf pieces were cut from the ng, va, and vg samples (eight pieces from each sample set). The leaf pieces (1 × 1 cm²) were temporarily held within agar gel and were cut using a vibrating blade microtome (50–100 μm thick, Leica VT1200S, Wetzlar, Germany). These sections were examined and photographed under a light microscope (Leica DM5500B, Wetzlar, Germany).

RNA data analysis and identification of DEGs

This study utilized mRNA-seq data and the full-length transcriptome generated in our previous study (accession number PRJNA1029356 in the NCBI database). To investigate leaf structure and identify DEGs, we selected eight leaf samples with the most abundant and darkest leaf spots, resulting in eight vg and eight va samples. We then randomly selected one non-spotted leaf sample from each sampling site, yielding eight ng samples. The mRNA data from all the ng, va and vg samples (24 samples in total) were mapped to the full-length transcriptome of C. forrestii using HISAT2 v 2.2.1 (Kim et al. 2019). For pathogen detection, RNA sequences with high similarity to microbial sequences were retained in the reference transcriptome. Differential gene expression was analyzed between the ng and va groups, as well as between va and vg. Aligned reads were quantified using StringTie v2.1.7 (Pertea et al. 2015) and DESeq2 v1.34.0 (Love et al. 2014). To minimize false positives, transcripts with fewer than five counts per million reads in at least eight samples were excluded. Functional annotation and GO/KEGG terms mapping of all potential genes were conducted using eggNOG-mapper v2.1.9 (Cantalapiedra et al. 2021). Genes with a fold change ≥ 1.5 and FDR < 0.05 were considered DEGs. GO/KEGG enrichment and visualization were performed using TBtools v1.120 (Chen et al. 2020). For detailed and concise identification, all transcripts were aligned with the NCBI NR/NT database and the Swissport protein knowledgebase (downloaded at 17/7/2023) via local BLAST v2.5.0.

Evaluation of ecological functions

Pollinator attraction

The floral scents of both morphs were collected on-site using dynamic headspace adsorption. Each newly opened flower was enclosed in a polyethylene terephthalate (PET) bag. One end of the bag was fitted with an activated carbon filter for air intake, and the other end was fitted with a volatile collection trap from Analytical Research Systems, containing 30 mg of Alltech Super-Q adsorbent material. Each flower was enclosed for approximately 3 h. Controls were established using empty PET bags. The adsorbed volatile samples were analyzed using automatic thermal desorption-gas chromatography-mass spectrometry (ATD-GC/MS) on a Shimadzu QP2020 NX system (Kyoto, Japan). To investigate whether leaf variegation in C. forrestii was associated with pollinator attraction, we recorded the number of flowers and fruits at all sites during 2022 and 2023. Additionally, field investigations were conducted at two sites in July 2023, where potential pollinators and the fruit-to-flower ratio were documented.

Photosynthetic parameters

The measurement of photosynthetic parameters was conducted at three sites. At each site, we randomly selected and measured six variegated and six non-variegated ramets. The va and vg areas, or ng areas of the sampled ramets, were each measured three times. Chlorophyll fluorescence parameters, specifically the maximum photochemical efficiency of photosystem II (Fv/Fm) and the quantum yield of photosystem II (ΦPSII), were measured using a pulse-modulated fluorescence system (LI-COR LI-6400, Lincoln, Nebraska, USA). Data collected from the ng, va and vg parts were statistically compared using Wilcoxon tests.

Herbivore damage

Leaves of C. forrestii were photographed in situ at eight sites in 2022, with a total of 622 leaves from 311 ramets surveyed (ranging from 40 to 160 leaves per site). In 2023, 526 leaves from 263 ramets across five sites were surveyed (ranging from 42 to 126 leaves per site). There was a significant difference in precipitation and air temperature between the two years, with 2023 being notably drier and hotter. This study utilized weather station data from the Yulong Snow Mountains Forest Dynamic Plot, providing an opportunity to compare the performance of the two leaf morphs under different climatic conditions.

Leaf areas were measured with ImageJ v1.53q (Abràmoff et al. 2004). To ensure consistency, all measurements were taken by a single individual. The sizes of variegated and non-variegated leaves were statistically compared using Wilcoxon tests. Both morphs were found to be attacked by several unidentified insects. Herbivore damage, characterized by actual foliage loss and a distinctive cutting edge, was distinguished from mechanical damage. This study focused solely on herbivore damage. A binomial mixed-effects model (GLMM) was employed for analysis, utilizing the R packages MuMIn v1.46.0 (Barton and Barton 2015) and glmm.hp v0.0-5 (Lai et al. 2022, 2023). In this model, the binary presence or absence of herbivore damage (0 or 1) served as the dependent variable, with the sampling site as the random factor. The presence or absence of leaf variegation (0 or 1) and leaf area were the independent variables, aligning with the method described by Toll (Toll 2023). The inclusion of leaf area allowed us to test the ‘Plant Apparency Hypothesis’, which posits that more apparent plants adaptively produce a greater quantity of chemical defenses as they are more easily detected by herbivores, while less apparent plants may invest less in defense. The areas of herbivore damage on both variegated and non-variegated leaves were statistically compared using Wilcoxon tests.

RESULTS

Leaf micromorphological features

Variegated leaves were significantly thicker than non-variegated ones (Fig. 3a). The thickness of the epidermal layers and upper mesophyll of both leaf morphs was very similar, but the spongy chlorenchyma cells (lower mesophyll) in the variegated leaves were generally looser and contained more air space. Trichomes, comprising 1–2 cells and about 50–150 μm in length, were found in the centers of the variegated areas. The upper epidermis and trichomes in these variegated areas were maroon or purple (Fig. 1a–f).

DEGs involved in leaf variegation

A total of 166 DEGs were identified when comparing non-spotted leaves to the green areas of variegated leaves, with 145 (88%) being upregulated in the green areas of the variegated leaves (Supplementary Table S1). Additionally, 263 DEGs were identified between the green and variegated areas of the spotted leaves, of which 171 (65%) were upregulated in the variegated samples (Supplementary Table S2). Transcripts resembling microbial sequences, such as transcript/8394 and transcript/12527, were not identified as DEGs. Subsequent functional analysis was performed using GO and KEGG enrichment analyses. The most significantly enriched KEGG pathways included ‘phenylpropanoid biosynthesis’, ‘flavonoid biosynthesis’ and ‘biosynthesis of other secondary metabolites’ (Supplementary Fig. S2).

DEGs between non-spotted leaves and the green areas of variegated leaves were enriched for several GO terms, with the most significant being ‘anthocyanin-containing compound metabolic process’, ‘flavonoid metabolic process’ and β-glucosidase activity (Fig. 2a). Among the anthocyanin-related DEGs, transcript/18129 and transcript/15197 exhibited high similarity to chalcone synthase (CHS) genes across various plants and were upregulated in the green areas of the variegated leaves. Additionally, transcript/8591, which is highly similar to BGLU22 (β-glucosidase 22 in Arabidopsis), was found to be upregulated (Fig. 1j, left column). Another major enriched GO term was ‘cell wall organization or biogenesis’. Most genes associated with this term were also upregulated, including transcript/24385 and transcript/31691, which showed strong similarity to xyloglucan endotransglucosylase/hydrolase (XTH23); and transcript/1222, which was closely related to the exocyst complex component (EXO). These findings are further illustrated in Fig. 1g and h, left column.

DEGs between the green and variegated parts of the spotted leaves were enriched for several flavonoid-related GO terms, including ‘caffeate O-methyltransferase activity” and “secondary metabolite biosynthetic process’ (Fig. 2b). Transcripts 10920 and 18201, which are highly similar to the caffeic acid O-methyltransferase (COMT) genes found in various plants, were upregulated in the variegated parts. Additionally, BGLU22-like transcripts 8591 and 9536 were downregulated in these areas (Fig. 1j, right column). DEGs associated with membrane activities, such as transcripts 22035 and 25161, which are very similar to glutathione S-transferase (GST), were upregulated (Fig. 1g and h, right column). Similarly, transcripts 12703 and 12983, which are very similar to multidrug and toxic compound extrusion (MATE) proteins, were also upregulated (Supplementary Table S2).

DEGs related to heat or drought responses were identified under the relevant GO term. Transcripts 18607 and 18779, both very similar to chitinase class I (CTL1), were upregulated in the green parts of the spotted leaves. Heat shock protein genes, including transcripts 4314, 11924, and 3291, were upregulated in the variegated parts. Furthermore, genes highly similar to BAG (Bcl-2-associated athanogene) were upregulated in the variegated samples (Fig. 1i). We also observed an increase in genes associated with the jasmonic acid (JA) signaling pathway. Specifically, PED1-like genes (transcripts 12675 and 13320) and the TIFY6B-like gene (transcript 24561) exhibited upregulation in the variegated samples (Fig. 1j).

Potential functions of leaf variegation

Both variegated and non-variegated individuals of C. forrestii are capable of flowering and fruiting. In 2022, 20 non-variegated ramets (10% of the total non-variegated ramets) flowered, with 4 (20%) producing fruit; 30 variegated ramets (27% of the total variegated ramets) flowered, of which 6 (20%) bore fruit. In 2023, 19 non-variegated ramets (9%) flowered, with only 1 (5.26%) setting fruit; in contrast, none of the variegated ramets (8% of the total) produced fruit (0%). Despite our observations, we did not identify any effective pollinators for C. forrestii. The primary volatile compounds contributing to the flower’s odor were acetoin, octanal, and benzaldehyde (as detailed in Supplementary Table S3), with no significant differences detected between the flower samples from variegated and non-variegated individuals.

The photosynthetic parameters of both leaf types were comparable. The average effective quantum yield of photosystem II (ΦPSII) values for the non-variegated (ng), variegated green (vg) and variegated anthocyanin-rich (va) parts were 0.695, 0.670 and 0.686, respectively. Similarly, the average maximum photochemical efficiency of photosystem II (Fv/Fm) values for the ng, vg and va parts were 0.763, 0.757 and 0.761, respectively. There was no significant difference in these parameters between the variegated and non-variegated leaves, as illustrated in Fig. 3b and c.

Variegated leaves were statistically larger than, or similar in size to, non-variegated leaves (overall: Fig. 3d; site-specific: Supplementary Fig. S3). The composition of the population varied across different sites, but the overall ratio of variegated leaves remained stable throughout our two-year observation period, averaging 36%–37% (Fig. 3g and h). Both morphs were attacked by several unidentified insects, including species from Chrysomelidae (Leaf Beetles). The herbivory rate, defined as the probability of being damaged by herbivores, was higher for non-variegated leaves compared to variegated leaves at both the site level (Fig. 3g) and the yearly level (Fig. 3h). In the GLMM analysis, leaf size was not found to be significantly correlated with the rate of herbivory. However, the herbivory rate was negatively associated with the presence of variegated spots, with significant P-values of 9.9e-4 in 2022 and 8.84e-11 in 2023 (Supplementary Data S1). The association was stronger in 2023, as indicated by a higher R² value in the model (0.15 compared to 0.04 in 2022, Supplementary Data S2).

In 2023, the herbivory rate for non-variegated leaves increased from 24% to 39%, while it remained almost unchanged for variegated leaves (12% to 13%, Fig. 3h). Overall, the leaf area consumed by herbivores did not significantly differ between 2022 and 2023 (P = 0.5). In 2022, the leaf area consumed by herbivores did not significantly differ between the two leaf morphs (P = 0.52, Fig. 3e). However, in 2023, the area of variegated leaf consumed was significantly smaller than that of the non-variegated leaf (P = 0.0036, Fig. 3f).

According to climate data from the Yulong Snow Mountains, the year 2023 was significantly drier and hotter than the historical average, while 2022 was a typical year in terms of precipitation and air temperature. The 2022/2023 growing season experienced 33% less precipitation compared to the 2021/2022 season, and air temperatures in April, May, and June of 2023 were 2.8 to 0.5°C higher than in the same period of the previous year (Supplementary Fig. S4).

DISCUSSION

Structural and molecular causes of leaf polymorphism

The variegated leaves of C. forrestii were found to be larger and thicker than their non-variegated counterparts, a difference that may be attributed to the differential expression of genes related to cell wall and membrane activities. One such gene, designated as Transcript/1222, exhibits similarities to the exocyst complex component (EXO). In Arabidopsis thaliana, EXO is recognized for its role in influencing cell expansion and leaf size (Schröder et al. 2009). The upregulation of a gene analogous to EXO in C. forrestii may, therefore, be a contributing factor to the enhanced size and thickness observed in the variegated leaves.

The adaxial epidermis layer and trichomes of the variegated leaves of C. forrestii exhibit a distinctive purple coloration, which is associated with anthocyanin production. The ‘anthocyanin-containing compound metabolic process’ is identified as the most enriched GO term among the DEGs when comparing variegated to non-variegated leaf samples. This suggests that the metabolic pathways leading to anthocyanin synthesis are upregulated in the variegated leaves, resulting in the characteristic purple coloration. Chalcone synthase (CHS) is recognized as a crucial enzyme in anthocyanin metabolism. In our study, transcripts similar to CHS, specifically transcript/15197 and transcript/18129, were found to be upregulated in the variegated leaf samples. Despite this, CHS and other downstream enzymes involved in anthocyanin synthesis do not show differential expression between the vg and va samples. Notably, unlike the va parts, the adaxial epidermis of the vg parts lacks purple-pigmented cells. One plausible explanation for the absence of anthocyanin in the vg areas could be the chemical instability of anthocyanins, potentially due to the degradative action of plant intrinsic enzymatic systems (Oren-Shamir 2009; Zhao et al. 2021).

The stability and light reflectivity of anthocyanins can be greatly enhanced by intermolecular co-pigmentation, such as with ferulic acid, a kind of phenolic acid (Eiro and Heinonen 2002). Caffeic acid O-methyltransferase (COMT) and related enzymes catalyze the synthesis of ferulic acid and the methylation of anthocyanins in various plant species (Giordano et al. 2016; Song et al. 2022; Wang et al. 2018). Transcript/10920 and transcript/18201 were found to be very similar to the COMT sequence and were upregulated in the va areas. The vacuolar transportation process is also known to further concentrate and stabilize anthocyanins. GST and MATE are two of the main anthocyanin transporters (Gomez et al. 2009, 2011; Zhao 2015), and their gene analogs (transcripts/22035, 25161, 12983 and 12706) were all upregulated in the va samples. In summary, the non-variegated leaves have a lower production of anthocyanin and other pigments. Meanwhile, in the variegated leaves, the vg areas produce higher amounts of anthocyanins which are colorless or are easily degraded; while in the va areas, the co-pigmentation and vacuolar transportation result in stable and highly visible pigment spots.

Leaf variegation in C. forrestii: physiology and function

Our results have ruled out a pathologic cause for the leaf variegation observed in C. forrestii. As previously mentioned, transcripts resembling microbe-like sequences, such as virus-like transcript/8394 and transcript/12527, were excluded from the DEGs analysis comparing variegated and non-variegated leaves (Supplementary Tables S1 and S2). Consequently, the leaf variegation in C. forrestii is likely a non-pathological physiological phenomenon.

We cannot definitively confirm or deny whether the leaf variegation in C. forrestii serves a function in pollinator attraction. However, it appears that leaf variegation may not be essential for pollination, given that both variegated and non-variegated ramets are capable of producing flowers and fruits. Moreover, variegation is not limited to the flowering stage but is also observed during the vegetative stage. We have not been able to discern any advantage in reproductive fitness between the two morphs, although we acknowledge the limitations imposed by our relatively small sample size of flowers and fruits.

The leaf variegation neither hampers nor amplifies the photosynthetic potential of C. forrestii. The ΦPSII and Fv/Fm values of ng, vg and va parts are statistically identical. It has been suggested that plant individuals with higher anthocyanin concentrations in their leaves may exhibit weaker photosynthesis performance under strong light conditions (Zhang et al. 2011). However, C. forrestii, being a typical understory plant, is seldom exposed to direct sunlight, suggesting that its two leaf morphs are likely to have similar productivity under natural conditions. Consequently, leaf variegation in C. forrestii is unlikely to be an adaptive trait for photosynthesis.

Leaf dimorphism: bet-hedging during climate fluctuation

Both morphs of C. forrestii are susceptible to attack by several unidentified insects, including species from the Chrysomelidae family (Leaf Beetles). Additionally, White-lipped Deer (Cervus albirostris) may cause damage to Cypripedium species (Fang Ye, personal communication). Our statistical analysis indicates that the probability of herbivory on the variegated leaves of C. forrestii is lower than that on non-variegated leaves across different sites (Fig. 3d) and years (Fig. 3e). This reduction in herbivory is not due to the variegated leaves being smaller or less accessible. In fact, the variegated leaves are statistically similar in size or larger than the non-variegated ones (Fig. 3c; Supplementary Fig. S3), and they are commonly found at all sites, representing approximately 36% to 50% of the leaf population (Fig. 3d and e).

An intriguing mystery thus remains: theoretically, if leaf variegation is dispensable and costly, it should be eliminated by natural selection. Conversely, if leaf variegation is of vital importance, or confers a significant advantage, most individuals would be likely to present the same leaf pattern. However, such intraspecific leaf variegation polymorphism is common among many plant groups, particularly in Begonia and species within the Gesneriaceae family. We hypothesize that the significance of leaf variegation may vary with certain environmental parameters, suggesting that the herbivory-avoidance function of leaf variegation could fluctuate under different conditions.

Field experiments (Castagneyrol et al. 2018; Field et al. 2020) and a paleontologically corrected meta-analysis (Hamann et al. 2021) have demonstrated that elevated temperatures and drought stress can significantly increase the incidence of herbivory. Our observations align with these findings, showing a higher incidence of damage to non-variegated leaves during the 2022/2023 season. Concurrently, the climatic conditions of the 2022/2023 season were markedly drier and hotter.

Our DEGs analysis supports a correlation between drought and leaf variegation. For instance, the upregulation of XTH23 has been shown to enhance the water-holding capacity of Populus, consequently improving its salt tolerance (Wang et al. 2022). This findings aligns with experiments in Arabidopsis, where upregulated XTH23 was found to promote lateral root development and enhance salt tolerance (Xu et al. 2020). Similarly, upregulated CTL1-like genes have been shown to contribute to heat and drought tolerance across various plant species (Hermans et al. 2010; Kesari et al. 2015; Kwon et al. 2007). Additionally, the upregulated DUF538-like gene may be involved in epidermal wax production (Li et al. 2019) or the regulation of stomata (Li et al. 2023), suggesting that variegated leaves may lose less water. Also, the upregulated BAG-like gene in variegated leaves may involve in stress responses and programmed cell death (Jiang et al. 2023). These gene expression patterns suggest an enhanced drought or other abiotic stress tolerance in the variegated C. forrestii plants.

Our DEGs analysis also supports a correlation between herbivore damage and leaf variegation, evidenced by an upregulation of genes associated with the jasmonic acid (JA) signaling pathway. Specifically, we observed upregulation in vg samples of PED1-like genes (transcripts/12675 and 13320), which encode the 3-ketoacyl-CoA thiolase (KAT) protein. This protein is essential for both local and systemic induction of JA biosynthesis, as demonstrated by study of various plants (Afitlhile et al. 2005; Castillo and Leon 2008; Castillo et al. 2004). We also observed the upregulation of TIFY6B-like gene (transcript/24561) in vg samples. This gene functions as a negative regulator of JA responses (Chung and Howe 2009). JA and its methylated form, methyl jasmonate (MeJA), are plant hormones and signaling molecules associated with plant defense responses, and have been shown to enhance the accumulation of flavonoids, anthocyanins and other secondary metabolites in plants (Loreti et al. 2008; Tamari et al. 1995). The upregulation of BGLU22 may also contribute to the plant’s defense mechanism. BGLU22 is involved in the targeted hydrolysis of scopolin, its natural substrate, a compound induced by various stressors such as NaCl and MeJA. Scopolin and its derivative, scopoletin, play roles in multiple aspects of plant defense (Ahn et al. 2009). Moreover, when cells are compromised by herbivore or pathogen attack, BGLU22 is released from the endoplasmic reticulum (ER) and interacts with jacalin-related lectins (JALs), leading to the formation of aggregates with increased glucosidase activity. These aggregates hydrolyze glucosides into toxic compounds (Yamada et al. 2011). In a related study on cotton, it was revealed that pigment spots induced by JA can reduce herbivore damage (Sun et al. 2024), further supporting the defensive role of JA-induced and pigmentation in plants.

From an ecological perspective, the defensive role of variegation can be realized via multiple strategies, including camouflage, aposematic signaling, insect or fungal damage mimicry, dazzle effects, lowering nutritional value for herbivores and other unknown mechanisms (Lev-Yadun 2014, 2019; Lev-Yadun and Niemelä 2017). In the case of C. forrestii, its leaf variegation may function as a warning signal indicating a higher content of anthocyanins and flavonoids; it could also be mimicking leaves that are damaged or infected. The herbivore deterrence function of anthocyanins and their flavonoids allies has been demonstrated by many studies (Karageorgou and Manetas 2006; Malone et al. 2009). Our DEGs results suggest that variegated leaves should contain higher concentrations of flavonoids. Herbivores are more likely to consume non-variegated leaves due to the absence of warning or mimicry markings. Under drought conditions, it is possible that herbivores consume less variegated C. forrestii because they have an overall increased intake of flavonoids from other stressed plants in the community, or a perhaps because they have a lower ability to detoxify these chemicals. It is highly possible that leaf variegation confers a greater advantage to C. forrestti under drought conditions.

Production of anthocyanins and other chemicals would bring extra metabolic costs. The bigger variegated leaves that are better defended are less susceptible to herbivore attacks and have a longer leaf lifespan to cover the cost. This supports the ‘cost–benefit hypothesis’. Non-variegated leaves may be more cost-efficient in non-drought years, whereas the variegated phenotype would be favored and established during long-lasting dry climates. No single leaf type has an overwhelming advantage, as no phenotype has been completely eliminated from the population. Similarly, no phenotype exhibits a significant disadvantage, since both can flower and bear fruit. Therefore, both phenotypes have been preserved by natural selection and coexist stably. The leaf variegation dimorphism may reflect two growth strategies and are a legacy of bet-hedging during ancient climate shifts.

Research deficiencies and prospects

This study, grounded in a two-year observational study of C. forrestii in its natural habitat, offers a revealing glimpse into the role of leaf variegation and the establishment of leaf polymorphism. However, long-term monitoring is anticipated to yield more compelling and comprehensive insights. Additionally, further field research is necessary to identify the effective pollinators and principal herbivores affecting C. forrestii. Moreover, the precise impact of leaf variegation and herbivory on the fitness of C. forrestii remains unclear.

We hypothesize that leaf variegation in C. forrestii may serve as a visual deterrent, signaling the presence of chemical defenses, or it may function as damage mimicry, suggesting a lower nutritional value. The defensive chemicals and nutrient content of the plant could vary significantly across different life stages or color morphs (Chen and Huang 2013). A metabolomic analysis could elucidate the trade-offs between defense mechanisms and growth in the distinct morphs.

No complete genome data is available for any species within the Cypripedioideae subfamily. The orchid genomes that have been published to date, such as those of Cymbidium ensifolium and Phalaenopsis equestris, pertain to other orchid subfamilies and exhibit low alignment rates with the transcriptome of Cypripedium species, approximately 60%. A well-assembled genome of Cypripedium specie would greatly enhance our understanding, particularly in unraveling the evolution and the development of leaf variegation.

The cultivation of C. forrestii and its closely related species is essential for ex-situ conservation efforts and for facilitating further research. The aseptic germination of C. forrestii seeds remains a technique under development. We are hopeful and confident that this challenge will be surmounted in the near future.

Supplementary Material

Supplementary material is available at Journal of Plant Ecology online.

Figure S1: Map of sampling sites.

Figure S2: KEGG enrichment result between (a) ng and vg samples; (b) vg and va samples.

Figure S3: Leaf size differences between variegated and non-variegated leaves of all 8 sampling sites.

Figure S4: The monthly average temperature and precipitation recorded at the Yulong Snow Mountain Station from July 2021 to June 2023.

Table S1: DEGs between ng and vg samples.

Table S2: DEGs between the vg and va samples.

Table S3: Volatile compounds from the flower odor.

Data S1: glmm model of the 2022 data.

Data S2: glmm model of the 2023 data.

Funding

This work was supported by Science & Technology Basic Resources Investigation Program of China for Survey and Germplasm Conservation of PSESP in Southwest China (2017FY100100); the Second Tibetan Plateau Scientific Expedition and Research Program (2019QZKK0502); the PSESP project of Yunnan Forestry and Grassland Bureau (2021SJ14X-09) and the project ‘Collection and Conservation of Plant Species with Extremely Small Populations of Polystichum glaciale and Cypripedium forrestii in Lijiang’ (2021SJ14X-11).

Acknowledgements

We thank LIU Yang (Kunming Institute of Botany, KIB), HE Zhixun (Lijiang Alpine Botanic Garden) and FANG Ye (Shangri-la Alpine Botanical Garden) for their assistance in in the field surveys. We appreciate the professional work of our KIB staff: the microtome preparation was aided by CHANG Wei and the flower odor was analyzed by SHI Yuqing. Photograph in Fig. 1c was provided by WANG Boqiang (student of KIB).

Conflict of interest statement. The authors declare that they have no conflict of interest.

References