Withering corolla remains on the calyx tube enhance reproductive success in Oxalis stricta L.

Abstract Several flower traits can affect plant reproductive fitness via pollinator attraction, herbivory defense, and thermal regulation of the pistil. In this study, we focus on thermal regulation of the pistil after flowering. We experimentally investigated the functional significance of the withering corollas that remain attached to the calyx tubes of Oxalis stricta L. We studied thermal regulation of the pistil by removing corollas and comparing the plants with and without corollas, under regulated dark and light periods, with an ambient temperature during the dark period lower than that during the light period. In plants lacking corollas, the pistil temperature was significantly lower than in control plants (with intact corollas) by approximately 2°C. Although fruit set in the corolla‐removed plants was not significantly different from that in control plants, the temperature threshold for 50% fruition in the corolla‐removed plants was significantly higher than that in the controls. Furthermore, the seed number, total seed weight, and single‐grain weight were significantly lower in the corolla‐removed plants than in control plants. The estimated annual number of reproductive cycles (from June to October), total seed number, and total seed weights were also lower in corolla‐removed plants. These findings indicate that the withering corolla remains play a role in thermoregulation of the pistil, and thereby enhance reproductive success. Our study is the first to validate one of the assumed ecological roles of the withering remains of plant corollas.

The significance of plant movements was originally examined by Darwin (1880), who conducted detailed studies of leaf opening and closure and hypothesized that their ultimate function was nocturnal thermal maintenance facilitated by a reduction in the leaf surface area when the leaves were closed. As Darwin indicated, diurnal movements could also be related to plant thermal regulation. The ecological role of temporary flower closure with regards to thermal regulation and pollen viability has been previously studied. Prokop et al. (2019), for example, showed that flower closure increased pollen viability and stigma receptivity in Crocus species via thermal regulation, and Liu et al. (2017) have reported that petal closure is beneficial with respect to reproductive success in Magnolia species.
In this study, we focused on a further potential thermoregulatory mechanism associated flower reproduction, that is, the role played by closed and withering corolla remains after flowers have permanently closed. In some species, when the flowers close permanently, their petals begin to wither a few days after blooming. In this regard, Van Doorn (2001) categorized the cessation of flower life into two phases, senescence (withering) and abscission, which tend to show consistent patterns within families. Although flower closure and petal senescence/abscission after pollination, which shorten flower life, have been well studied (van Doorn & Kamdee, 2014), the research has focused primarily on proximate factors, such as the physiological aspects. Recently, however, Kyogoku, Kataoka, and Kondoh (2019) reported that the flowers of a diploid Taraxacum species rapidly close their heads in response to the landing of pollen, and also discuss the ecological aspects of this closure.
During the process of permanent flower closure, some species retain their withering corollas longer than others. However, the ecological function of the withered petals or corolla remains, and the reason why they remain attached to the calyx tube after pollination and during the fruiting period, have yet to be examined in detail.
As flowers bear important reproductive organs, such as pistils and stamens, they are assumed to have evolved certain adaptive traits (Abdusalam & Tan, 2014), such as protection from airborne pathogens and frugivorous animals (van Doorn & van Meeteren, 2003;Wu & Yahara, 2017). Therefore, in this study, we sought to investigate whether the withering remains of corollas that are retained on the calyx tube have any ecological significance.
We hypothesized that the withering corolla remains have a universal function of maintaining constant pistil and ovule temperatures, which contributes to reducing the level of seed deterioration.
We tested this hypothesis by performing experiments on Oxalis stricta L. (Oxalidaceae) to examine the effects on seed production of removing gamopetalous corollas from flowers.

| Plant material
We used O. stricta, a perennial herb that grows in circles in grasslands, as our experimental plant. It is widely distributed from temperate to subarctic zones in East Asia and eastern North America, and naturalized in Europe and Africa, flowering in June-October (Ohashi, Kadota, Murata, Yonekura, & Kihara, 2016b). Co., Ltd.) at 25°C with a fluorescent light (FL15N; Panasonic Co., Ltd.). All flowering buds except one were picked from each plant before blooming, because if a plant has both damaged flowers (such as those with the corolla removed) and intact flowers, damaged flowers are naturally aborted in some plant species (Stephenson, 1981).

| Laboratory experiments
Despite the plants favoring self-pollination with the abovemen- Throughout the experiment, the ambient temperature in the dark period was set to four different levels, that is, 23, 18, 13, and 8°C, using the abovementioned incubator. During the light period, the temperature was set to 25°C for all plants, and the plants were kept under a photoperiod of 12-hr light and 12-hr dark, starting at 19.00 and 07.00, respectively, to eliminate the effects of the circadian rhythm.
We measured the temperature of each pistil using an infrared thermometer (IR-308; Custom Co., Ltd.) after every 4 hr for 24 consecutive hours (at 00.00, 04.00, 08.00, 12.00, 16.00, and 20.00, of which 08.00, 12.00, and 16.00 fall in the dark period). Pistil temperatures were measured, and the measurements were repeated twice in the middle of each flowering period. The closed corollas were temporarily opened with tweezers and their temperatures were measured using a thermometer.
We recorded seed setting in all the plants and calculated the total seed number and weight per flower, using a Mettler balance (METTLER AE260; Mettler Toledo Co., Ltd.

| Statistical analyses
We used R 3.5.2 for Mac OSX (R Core Team, 2018) for all statistical analyses. Differences in ambient and pistil temperatures were tested using Welch's t test. Differences in fruit sets between the two corolla removal states were tested using the prop.test function in R, which tested the null hypothesis that the proportions (probabilities of success) in several groups were the same based on the Pearson's chi-squared test, and the threshold temperature for 50% fruition was calculated using a logistic regression with a binomial distribution.
Differences in fruition rate, seed number and weight were analyzed using a generalized linear model (GLM) and likelihood ratio test with the research site (two sites), night temperature (four levels), and the corolla removal treatment (removed and intact) as the explanatory variables. The fruition rate, seed number, and seed weight was modeled using binomial distribution, Poisson distribution, and Gaussian distribution, respectively. Afterwards, differences in seed number and weight between corolla movement treatments were tested using pairwise Welch's t tests.  in corolla-removed and control plants, respectively. The mean corolla-fall-off periods were 12.5 days after blooming (SD = 1.55) and 7.8 days before seed maturity (SD = 4.29) in control plants. One of the corolla-removed plants and its control pair were excluded from the analyses as the corolla-removed plant, at the end of its fruition, was damaged by a serious earthquake (i.e., 2018 Hokkaido Eastern Iburi Earthquake).

| RE SULTS
Pistil temperatures during the dark period were significantly lower in the corolla-removed plants than those in the control plants (t test, p < .001; Figure 2 We then tested the difference between the removal conditions by pairwise t-test. The mean total seed number per plant was 8.44 (n = 9, SD = 3.36) and 21.2 (n = 15, SD = 9.54), and the mean total seed weight per plant was 0.0013 g (SD = 0.00054) and 0.0041 g (SD = 0.0018) in corolla-removed and control plants, respectively.
The seed number in control plants was significantly higher than that in corolla-removed plants (p < .001; Table 1). The total and single seed weights in control plants were significantly higher than those in corolla-removed plants as shown by pairwise Welch's t test (p < .001 and p < .05, respectively; Table 1).
Td in June − October was calculated from the meteorological data of 29 years (mean = 44,082 ADH, SD = 2,788; Figure 4a). The mean effective temperature demands, Te and Te', were 9,064.00 and 8,029.60 ADH, respectively, and Te was significantly higher than Te' as determined by pairwise Welch's t test (p < .001). The mean number of reproductive cycles calculated using Td was 9.72 (8.71-10.95; Figure 4b) and 10.98 (9.83-12.36; Figure 4c) in corolla-removed and control plants, respectively, and the number of reproductive cycles was significantly higher in control plants than those in the corolla-removed ones, as shown by pairwise Welch's t test (p < .001).
The estimated total seed number and weight during one blooming season in control plants were both significantly higher than those in corolla-removed plants as determined by pairwise Welch's t test (p < .001, p < .001, Table 1).

F I G U R E 2
Difference between the pistil temperature of Oxalis stricta and ambient temperature for each nocturnal temperature level of the incubator. The difference between corolla-removed and control individuals was significant by Welch's t test. ***p < .001

| D ISCUSS I ON
As hypothesized, we found that pistil temperatures in corolla-removed plants were significantly lower than those in control plants, indicating that withering corolla remains are effective in maintaining the pistil temperature. This suggests that the function of the retained corolla is to prevent pistil exposure to cold ambient air.
This effect is more remarkable at lower ambient temperatures, when exposure could decrease pistil temperature. Therefore, this suggests corolla remains are beneficial and increase reproductive fitness. Regarding the cost of retaining the withered corolla, the cost of retaining dying or dead tissue is considered to be lower than that required for the maintenance of living tissue such as petals (Ashman & Schoen, 1997;Franchi, Nepi, & Pacini, 2014;He, Duan, Liu, & Smith, 2005).
As all other organs were intact in the corolla-removed flowers and no additional flowers receiving reproductive resources were left on the plants, selective abortion was not activated in the corolla-removed plants (Stephenson, 1981). Thus, fruition rate did not significantly differ between the corolla-removed and control plants. However, the threshold temperature for 50% fruition in the corolla-removed plants was higher than that in control plants.
The seed weight per grain and seed number per plant in corolla-removed plants were lower than those in the control. This indicates the possible role of the withering corolla remains in the production of more and bulkier seeds by maintaining the pistil temperature, which is the most significant function of the withering corolla remains on flowers. Our results are consistent with preceding studies such as Wada (1998), Zhang, Ai, Yu, Wang, andLi (2010), and Liu et al. (2017), which reported that thermal keeping by heliotropism or petal closing enhance seed production.
Trade-offs between the seed number and grain weight would suggest either more lighter seeds or fewer heavier seeds.
Contrastingly, our results showed higher numbers of heavier seeds in the control plants. As the fruit set did not differ be- TA B L E 1 Results of petal_removal experiment for Oxalis stricta. Seed number and seed weight (g) were measured and weight per grain is calculated. The number of reproduction, total seed number, and total seed weight were estimated using meteorological data.
caused by differences in pollen-tube growth or in maturation after fertilization.
We had expected the seed maturation in corolla-removed plants to take longer than that in the control as the recorded Te was significantly higher than Te'. Thus, the potential number of reproductive cycles per year in corolla-removed plants was estimated to be lower than that in the control. Consequently, the total number of seeds per year in corolla-removed plants was also lower than that in the control plants.
Our results were derived from data of 64 assessed individuals of one plant species. However, the corolla removal treatment certainly affected plant reproductive fitness regarding the seed number, total seed weight, and single-grain weight. Other plant taxa, such as

ACK N OWLED G M ENT
We gratefully thank Dr. K Horie, the manager of Asahikawa City Northern Wild Plants Garden, who helped us to collect O. stricta individuals.

CO N FLI C T O F I NTE R E S T
The authors have declared that no competing interest exists.

AUTH O R CO NTR I B UTI O N S
A.I. conceived the study, conducted analyses, and wrote the manuscript. K.H. conducted experiments.

DATA AVA I L A B I L I T Y S TAT E M E N T
All relevant data are within the paper and its Supporting Information files.