Introduction

Underwater light availability is one of the most important environmental factors affecting the growth, morphology, species composition and distribution of submersed macrophytes in lakes1. For many submersed macrophyte species growing on different light regimes required physiological adaptations and patterns of it might be species-specific2. In lakes, low light stress on submersed macrophytes could be induced by many biotic and abiotic factors such as resuspension of soft surface sediment, increases in periphyton and phytoplankton, and shading of neighboring plants3,4. Eutrophication decreases the water transparency and combined with water level fluctuations, it not only affects the vertical distribution of underwater light intensity but also the amount of light reaching the sediment shifting the relative importance of nutrients versus light availability in affecting growth, distribution and species interaction of submersed macrophytes5,6. Based on the morphological characteristics of submersed macrophytes, Chambers7 classified them into four types of growth forms: bottom dweller, rosette, erect, and canopy former. Each form prefers different habitats relative to nutrient status and light regime, indicating species-specific adaptive strategies to the various environments. Several studies have examined the physiological and morphological responses of submersed macrophytes to various light regimes6,8,9,10. In case of low light availability in water column, submersed macrophytes might adopt one of two distinct strategies – elongation of shoot length towards water surface to alleviate low light stress and/or enhancing low light tolerance by photosynthetic adjustments. Few studies, however, have linked the two low light adaptive strategies to the distribution of submersed macrophytes along various light gradients and evaluated their relative importance.

Rapid light curve (RLC) estimates the relative electronic transport rate (rETR) as a function of photosynthetic active radiation (PAR) by using modulated chlorophyll fluorescence technology, which provides information on the temporal acclimated state of photosynthesis. The minimum saturating irradiance (Ek), the initial slope of RLC (α) at extremely low light region, and the maximum relative electron transport rate (ETRm) inferred by the RLC reflect the ability of plants to tolerate high light, the light use efficiency and the maximum photosynthetic rate of the plants, respectively11. Morphological parameters, relative growth rate (RGR) and clonal reproduction provide information on the ecological adaptation of submersed macrophytes to various light regimes, because the survival, fulfillment of life history and population expansion of the plants require extra photosynthetic carbohydrate supply than those using for leaf maintenance alone12.

In lakes, two common submersed macrophytes, Potamogeton maackianus A. Benn. (canopy former)7 and Vallisneria natans (Lour.) H. Hara (rosette) grow well in waters with medium turbidity. They can both be dominant in such an environment despite their different growth forms13, indicating that their low light adaptive strategies differ significantly. In this study, P. maackianus and V. natans were cultured at various experimental light regimes with the aim to explore their low light adaptive strategies by measuring the RLC, morphological parameters, RGR and clonal reproduction and relating our findings to their abundance in two Chinese lakes. Specifically, we hypothesized that 1) P. maackianus and V. natans have different low light adaptive strategies as they differ in growth form, potentially leading to a trade-off between physiological and morphological responses to low light stress; 2) P. maackianus as a canopy former might alleviate the low light stress by elongating its shoot length towards the water surface, while V. natans, being a rosette with low shoot elongation capacity rather, relies on photosynthetic adjustments to cope with the low light stress; 3) photosynthetic adjustments would become more important in determining plant abundance in deep water due to the lower carbon requirement compared with shoot elongation.

Materials and Methods

Experiment design and layout

The experiment was conducted from 20 April to 20 July 2015 at the Donghu Experimental Station of Lake Ecosystem (30°32′53.41″N, 114°21′15.63″E) located in Wuhan City, China. In the experiment, P. maackianus and V. natans were cultured in 32 outdoor aquaria (Length: 50 cm; Width: 50 cm; Height: 80 cm) placed in a shelter covered by black nylon nets with various thickness, resulting in four light regimes – 2.8% (I1), 7.1% (I2), 17.1% (I3), and 39.5% (I4) of the air light intensities reaching the aquaria.

At the beginning of the experiment, three healthy seedlings of V. natans (31.72 ± 8.72 cm height) were planted evenly in two plastic boxes (Length: 30 cm; Width: 17 cm; Height: 12 cm) containing 10 cm sediment collected from Lake Donghu and incubated in an aquarium filled with 70 cm water. As to P. maackianus, one healthy shoot (25.0 ± 3.1 cm height) was planted in a plastic cup (Diameter: 6.5 cm; Height: 9.8 cm) containing 9 cm sediment, and 30 cups were incubated in an aquarium filled with 70 cm water. The water used was a mixture of 70% purified water and 30% water from Lake Donghu. Four replicate treatments were carried out for each of the four light regimes (Fig. 1).

Figure 1: Experimental layout for the Vallisneria natans (Vn) and Potamogeton maackianus (Pm) cultivation experiment.
figure 1

(a) Aquaria (50 × 50 × 80 cm) filled with 70 cm water for cultivation of macrophytes. (b) The P. maackianus tanks each contained 30 cups with one P. maackianus individual in each cup, the V. natans tanks contained two boxes with three V. natans individuals in each.

Full size image

During the experiment, chlorophyll a (Chl-a), pH, temperature (T), and PAR was determined in the overlying water every two weeks according to the methods described by Clesceri et al.14. pH and temperature (T) were measured using a multifunctional YSI meter (Yellow Springs Instruments, Ohio, US) and PAR was measured by a Li-COR UWQ-192S sensor coupled with a Li-1400 data logger (Li-Cor, Lincoln, NE, USA) at 10:00–12:00 a.m. The mean and range (in the parentheses) were 0.48 (0.14–1.46) mg L−1 for total nitrogen (TN), 0.014 (0–0.075) mg L−1 for NH4-N, 0.21 (0–0.87) mg L−1 for NO3–N, 0.03 (0.01–0.08) mg L−1 for total phosphorus (TP), 0.011 (0.0006–0.069) mg L−1 for PO4-P and 2.48 (0–11.16) μg L−1 for Chl-a in the water. The mean and range (in the parentheses) of pH and T were 8.63 (7.42–10.24) and 25.7 (21.1–29.4) °C, respectively. The pH value of lake water nearby our experimental site was 9.08. The mean and range (in the parentheses) of PAR in the water surface in the I1 to I4 light regimes were 22.2 (9.13–43.72), 68.10 (25.57–136.45), 150.30 (60.15–294.77), and 348.89 (131.14–635.98) μmol m−2s−1, respectively. The contents of TN, NH4-N, NO3–N, TP, and PO4-P in the sediment pore water were measured at the beginning and at the end of the experiment following the method of Clesceri et al.14, and the mean contents were 1.68, 1.09, 0.34, 0.22, and 0.084 mg L−1, respectively.

Determination of photosynthetic, morphological and growth parameters of the experimental plants

Ramet numbers of V. natans and P. maackianus were counted on the 30th, 60th, and 90th day. The shoot length of P. maackianus, leaf length of V. natans and the branch numbers of P. maackianus were measured on day 60 of the experiment. At the end of the experiment, all plants were harvested, gently washed and dried with tissue paper and then weighted. The RGR of the plants was calculated by the formula: RGR = ln (M2/M1)/dt, where M2 and M1 were plant fresh weight at the end and beginning of the experiment, respectively and dt was the duration (days) of the experimental period.

The RLC of V. natans and P. maackianus leaves was measured by a WATER-PAM fluorometer (WALZ Company, Germany) using WinControl-3 software (v. 3.23), operating at 9 light intensity gradients (35, 79, 119, 175, 268, 396, 605, 903, and 1282 μmol m−2s−1), each light gradient lasting for 10 seconds. The RLC measurements were conducted at 3:00–6:00 p.m. on a sunny day (June 24) for 2 to 5 mature plant leaves in each aquarium. V. natans leaves were measured at 5 cm position from the leaf apex and the P. maackianus leaves were measured in center of leaf concave. RLC parameters were calculated by a fitting formula of White and Critchley15.

Field investigation of plants abundance in response to underwater PAR

Field investigations of the abundance of P. maackianus and V. natans were carried out in Xukou Bay (ca. 8.4 km2; max. water depth: 2 m) of Lake Taihu (2445 km2) in summer 2014 and in whole Lake Erhai (250 km2; max. water depth: 21.0 m) in summer 2015 where these two species are the dominant submersed macrophytes13. Lakes Taihu, Erhai, and Donghu (the experimental site) are located in the eastern, middle and south-western parts of China, respectively and thus belong to subtropical climate zone, and all three lakes have undergone eutrophication contributing to the decline of submersed vegetation during the past three decades16. The concentrations of TP and TN and Secchi transparency in the water column were 0.024 mg L−1, 0.32 mg L−1, and 112 cm in Xukou Bay of Lake Taihu and 0.03 mg L−1, 0.75 mg L−1, and 153 cm in Lake Erhai, respectively.

In the field investigations, submersed macrophytes were collected by an underwater reaping hook (covering a bottom surface area of 0.20 m2) in triplicate at each site. In Lake Erhai, the sampling sites were set at intervals of 0.5-m water depths along transects starting from the shore to the maximum depth of plant occurrence. A total of 1200 samples were collected from 87 transects uniformly distributed around the lake shore. At transects with a steep lake bottom, submersed macrophytes were sampled at fewer sites. In the Xukou Bay of Lake Taihu, the sampling sites were set at intervals of 20-m distance along transects starting from the shore to the deepest depths where the plants occurred. A total of 108 samples were collected from 5 transects uniformly distributed around the lakeshore. P. maackianus and V. natans were separated from the collected macrophytes, gently washed, and weighed to determine the fresh biomass (FW) at each sampling site.

We measured the PAR in each sampling site at several water depth (0, 0.5, 1, 1.5, 2, 2.5 … m) for five repetition according to the actual sampling site water depth. The light extinction coefficient (K) of the water column was calculated based on the equation: K = (lnI1–lnI2)/(d2–d1), where d stands for water depth and subscript stands for water depth order: 1 is the lower position and 2 is the deeper position. I1 and I2 is PAR at water depth d1 and d2, respectively17. Then the K in the sediment (Ks) was obtained by fitting K in each water depth (0, 0.5, 1, 1.5, 2, 2.5 … m). Therefore, the light reaching the sediment in the field can be calculated according to the PAR in water surface (0 m), Ks and water depth. Finally, we calculated the light transmittance (PAR in sediment/PAR in water surface) and classified them into four groups of RI1, RI2, RI3 and RI4, representing groups with light transmittance values nearest to the experimental I1, I2, I3 and I4, respectively. We excluded the values bang in the middle of two treatments. The mean and range (in the parentheses) of actual water depth ranges of two species in the RI1 to RI4 in Lake Erhai were 4.78 (4.20–5.50), 2.97 (2.70–3.30), 2.14 (1.60–2.60), 0.72 (0.30–1.00) m, respectively. The mean and range (in the parentheses) of actual water depth ranges of two species in the RI1 to RI4 in Xukou Bay of Lake Taihu were 1.84 (1.80–1.87), 1.77 (1.35–1.95), 1.48 (0.90–1.85), 1.36 (1.10–1.60) m, respectively.

Statistical analysis

Statistical analysis was carried out using IBM SPSS Statistics 19.0. One-way analysis of variance (ANOVA) was conducted to determine the statistical significance for each variable between the treatments. ANOVA results were considered significant at P < 0.05 and multiple mean comparisons were performed by Duncan’s test (at 0.05 significance level) to identify differences between the treatments. Before performing one-way ANOVA, all data were tested for normality and homogeneity. Non-normal data were Sqrt-transformed to obtain normality. Pearson’s correlations analysis was used to test for relationships between RGR and fluorescence parameters.

Results

The photosynthetic RLC response of V. natans and P. maackianus to the various light regimes

The photosynthetic RLCs and their parameters (α, Ek and ETRm) of V. natans and P. maackianus were significantly affected by the experimental light regimes (Figs 2 and 3, F = 5.42, 6.65 and 8.76 for α, Ek and ETRm, respectively, and p < 0.05 for all). According to the RLCs, the rETR reached its maximum value at 333, 424, 477, and 494 μmol m−2 s−1 light intensity for V. natans (I1 to I4 light regimes) and 353, 521, 568, and 557 μmol m−2 s−1 for P. maackianus (Fig. 2). The Ek values of V. natans decreased gradually with decreasing light regime, with average Ek values of 155.4, 122.1, 109.1, and 81.3 μmol m−2s−1 in the I4 to I1 light regimes, respectively. The Ek values of P. maackianus decreased marginally when light availability decreased from the I4 to I2 regimes, and then dropped significantly when the light availability decreased from the I2 to I1 regimes, with average Ek values of 184.5, 189.8, 175.0, and 131.9 μmol m−2s−1 at the I4 to I1 regimes, respectively. The Ek values of P. maackianus were higher than those of V. natans in each light regime (Fig. 3A). The ETRm values of V. natans and P. maackianus showed a unimodal response to the light gradients and reached the highest values in the I3 light regime. The average ETRm values of P. maackianus were 29.9, 50.2, 53.7, and 44.1 μmol m−2s−1 in the I1 to I4 light regimes, respectively, and higher than the 27.8, 33.7, 39.9, and 35.6 μmol m−2s−1 of V. natans, particularly for the I2 and I3 light regimes (Fig. 3B). Both species had increased α when the light availability decreased, with the highest value occurring in the I1 treatment; however, the average α value in the I1, I2, and I3 treatments of V. natans increased by 36.5% as compared to I4 treatment and the average α value in the I1, I2, and I3 treatments of P. maackianus increased by only 26.3% as compared to I4 treatment (Fig. 3C).

Figure 2: Effects of the treatments on rapid light curves (according to the formula rETR = Pm *(1 −EXP(−α* PAR/Pm)) *EXP(−β* PAR/Pm), where α, β, rETR, and Pm are the initial slope of the curve, photo-inhibition parameters, the relative electron transport rate, and the maximum potential relative electron transfer rate without photo-inhibition, respectively) of V. natans and P. maackianus on day 65 of the experiment.
figure 2

The I1, I2, I3 and I4 stand for 2.8%, 7.1%, 17.1% and 39.5% ambient light, respectively.

Full size image
Figure 3: Effects of the treatments on fluorescence parameters (Ek, α, and ETRm) of rapid light curves for V. natans and P. maackianus on day 65 of the experiment (mean ± SE, n = 4).
figure 3

The I1, I2, I3 and I4 stand for 2.8%, 7.1%, 17.1% and 39.5% ambient light, respectively. The A, B and C indicate parameters Ek, ETRm and α respectively. Different letters indicate Duncan’s test at the 0.05 significance level.

Full size image

Morphological and growth responses of V. natans and P. maackianus to the various light regimes

The decreasing light regimes influenced the growth of V. natans and P. maackianus to a different extent. The RGR of two species were significantly affected by the experimental light regimes (F = 15.78, p < 0.05). The RGR of V. natans decreased by 90.2%, 40.3%, and 20.2% in the I1 to I3 light regimes, respectively, compared with the I4 light regime. The RGR of P. maackianus did not change significantly in the I3 light regime, exhibited a slight 21.0% reduction in the I2 light regime and declined dramatically (RGR = −0.0018) in the I1 light regime compared with the I4 light regime. The RGR of V. natans was lower than that of P. maackianus in the I2 to I4 light regimes but higher than that of P. maackianus in the I1 light regime (Fig. 4).

Figure 4: Effects of the treatments on the relative growth rate (RGR, mean ± SE, n = 4) of V. natans and P. maackianus on day 90 of the experiment.
figure 4

The I1, I2, I3 and I4 stand for 2.8%, 7.1%, 17.1% and 39.5% ambient light, respectively. Different letters indicate Duncan’s test at the 0.05 significance level.

Full size image

The leaf length of V. natans and shoot length of P. maackianus were significantly affected by the experimental light regimes (F = 14.29, p < 0.05). The leaf length of V. natans and shoot length of P. maackianus showed a unimodal response to the increasing light gradients. The average leaf lengths of V. natans were highest (65.0 cm) in the I2 light regime, intermediate (mean: 49.3 cm) in the I1 and I3 light regimes and lowest (31.5 cm) in the I4 light regime. The average shoot length of P. maackianus was highest (69.5 cm) in the I3 light regime and then decreased to 45.8 cm, 28.9 cm, and 19.2 cm in the I4, I2, and I1 light regimes. The leaf length of V. natans was about twice as high as those of shoot length of P. maackianus in the I1 and I2 light regimes but 26.6% and 31.3% lower than those of P. maackianus in the I3 and I4 light regimes, respectively (Fig. 5A). The branch numbers of P. maackianus was significantly affected by the experimental light regimes (F = 26.75, p < 0.05). The branch numbers of P. maackianus increased substantially with increasing light regimes and were 3-, 10.5-, and 18-fold higher in the I2 to I4 light regimes than in the I1 light regime, respectively (Fig. 5B).

Figure 5: Effects of the treatments on plant height (mean ± SE, n = 4) of V. natans and P. maackianus and branch number (mean ± SE, n = 4) of P. maackianus on day 60 of the experiment.
figure 5

The I1, I2, I3 and I4 stand for 2.8%, 7.1%, 17.1% and 39.5% ambient light, respectively. Plant height means average maximum leaf length of each individual for V. natans and average perpendicular distance from each branch apex to stem bottom for P. maackianus in each aquarium. Branch number were the number of shoots except the highest main branch of plant in a cup. The A indicates plant height of two species and B indicates branch number of P. maackianus. Different letters indicate Duncan’s test at the 0.05 significance level.

Full size image

The ramet number increased substantially with increasing light intensity during the experimental period for V. natans (F = 70.53, p < 0.05) in the I1 to I4 light regimes and for P. maackianus (F = 18.02, p < 0.05) in the I3 and I4 light regimes. At the end of the experiment, V. natans had average 0.25, 5.5, 23.3, and 75.8 ramets in the I1 to I4 light regimes, respectively (Fig. 6A). P. maackianus had average 2.4 and 5 ramets in the I3 and I4 light regimes, respectively, and no ramets in the I1 and I2 light regimes (Fig. 6B).

Figure 6: Effects of the treatments on the ramet number (mean ± SE, n = 4) of V. natans and P. maackianus on day 30, 60 and 90 of the experiment.
figure 6

The I1, I2, I3 and I4 stand for 2.8%, 7.1%, 17.1% and 39.5% ambient light, respectively. The A and B indicate ramet number of V. natans and P. maackianus, respectively. Different letters indicate Duncan’s test at the 0.05 significance level.

Full size image

Abundance of V. natans and P. maackianus along light gradients in lakes

The biomass of V. natans and P. maackianus showed a unimodal response to the increasing light gradients in Lake Erhai and the Xukou Bay of Lake Taihu. The biomass of two species were significantly affected by the light regimes in Lake Erhai (F = 55.44, p < 0.05). The average biomass density of V. natans was highest (2523 g m−2FW) in the RI2 light regime, intermediate (mean: 1570 g m−2FW) in the RI1 and RI3 light regimes and lowest (629 g m−2FW) in the RI4 light regime in Lake Erhai. The average biomass density of P. maackianus was highest (6428 g m−2FW) in the RI2 light regime, intermediate (3607 g m−2FW) in the RI3 light regime and lowest (mean: 1405 g m−2FW) in the RI1 and RI4 light regimes in Lake Erhai. The average biomass density of P. maackianus was 60.7%, 51.8%, and 57.5% higher than the biomass of V. natans in the RI2 to RI4 light regimes and comparable to that of V. natans in the RI1 light regime in Lake Erhai (Fig. 7A).

Figure 7: Field investigations on biomass of V. natans and P. maackianus in Xukou Bay of Lake Taihu in summer 2014 and Lake Erhai in summer 2015.
figure 7

The RI1, RI2, RI3 and RI4 represent light availability reaching the sediment and were equivalent to the experimental I1, I2, I3 and I4 light intensities, respectively. Different letters indicate Duncan’s test at the 0.05 significance level.

Full size image

The biomass of two species were significantly affected by the light regimes in Lake Taihu (F = 8.92, p < 0.05). In Xukou Bay of Lake Taihu, the average biomass density of V. natans was highest (mean: 1398 g m−2FW) in the RI2 and RI3 light regimes, intermediate (688 g m−2FW) in the RI1 light regime and lowest (75 g m−2FW) in the RI4 light regime in Xukou Bay of Lake Taihu. The average biomass density of P. maackianus was highest (6590 g m−2FW) in the RI3 light regime and then decreased to 3995 g m−2FW and 1746 g m−2FW in the RI4 and RI2 light regimes, respectively; no plants were found in the RI1 light regime. The average biomass density of P. maackianus was 4.6- and 53-fold higher than those of V. natans in the RI3 and RI4 light regimes, respectively, and comparable to that of V. natans in the RI2 light regime (Fig. 7B).

Relationships between RGR and fluorescence parameters (α, Ek and ETRm)

Pearson’s correlations analysis between RGR and fluorescence parameters showed a significant positive relationship between RGR and α and Ek of V. natans and between RGR and α, Ek, and ETRm of P. maackianus. For V. natans, RGR exhibited highest correlation with α. For P. maackianus, RGR exhibited highest correlation with Ek (Table 1).

Table 1 Pearson’s correlation coefficient matrix of RGR and fluorescence parameters of V. natans and P. maackianus (n = 16).
Full size table

Discussion

The field investigations revealed that P. maackianus had higher biomass density than V. natans when light intensity was high (RI2, RI3, and RI4), while V. natans had higher biomass density in the low light stress environment (RI1). The changes in abundances of P. maackianus vs. V. natans along the underwater light gradients imply that the plants exhibit different strategies in their adaptations to the various light habitats.

The significant high positive correlation between RGR and α of V. natans suggests that the plant adjusts its physiological responses to the low light habitats (deep water) when enhancing its light capturing ability as alleviation of low light stress by leaf elongation towards the water surface become difficult for a rosette submersed macrophyte. By contrast, the significant high positive correlation between the RGR and Ek of P. maackianus suggests its increased photosynthetic production in the high light habitats (shallow water). The photosynthetically saturating light intensity promote shoot elongation and increased branch numbers of this canopy-forming submersed macrophyte. These results are in agreement with our first and second hypotheses.

The initial slope of RLC (α) reflects the light capturing ability of plant leaves11, and submersed macrophytes growing under low light stress usually have increased α value18,19,20. In the experiments with various light regimes, the higher proportional increases in α of V. natans indicated its better adaptation than P. maackianus to the decreased light availability. Both species tended to exhibit increased Ek and ETRm with increasing light intensity, which is consistent with the results of several other studies11,21, However, the lower Ek and ETRm of V. natans imply that relatively low light intensity may impact the photosynthetic light saturation, likely exposing the plant to excess light energy at high light intensity18,20. The higher shoot length and high branch number in the relatively high light environment induced the canopy formation of P. maackianus leading to light harvesting advantage of P. maackianus over V. natans8,22. However, the continuous spectrum of increasing Ek relative to the light intensities of V. natans indicated higher plasticity in its photosynthetic physiological responses to the various light regimes than that of P. maackianus whose Ek and ETRm were saturated at the relatively low light regimes (I2). This may contribute to the more rapid decline in P. maackianus than in V. natans abundance at the low light regimes in Lake Erhai and Xukou Bay.

In the experiments, the leaf lengths of V. natans and shoot length of P. maackianus showed a unimodal response to the decreasing light gradients, indicating that the plants adapted to the moderate but not to the extremely low light stress via leaf or shoot elongation alone. In the I1 light regime, the growth of P. maackianus was suppressed severely and its RGR became negative, meaning the plant hard to survive for prolong time, while V. natans had positive RGR and survived likely due to its photosynthetic adaptation to low light stress, reflected by the lower Ek of V. natans than that of P. maackianus. Similar to the leaf or shoot length responses observed in the experiments, in the field investigations the biomass of V. natans and P. maackianus exhibited a unimodal response to the decreasing light gradients, implying a close relationship between biomass and leaf or shoot length in response to various light gradients. In the lakes, increasing water depth not only supplies more space for plant growth but also induces leaf or shoot elongation due to the decreased underwater light availability, leading to linear responses in biomass to increasing water depths until leaf or shoot elongation is limited by the severe low light stress in deep water. As longer and slimmer shoots of submersed macrophyte are more prone to be damaged by hydraulic forces23, the photosynthetic adjustments combined with leaf elongation of V. natans may have promoted its survival under the severe low light stress conditions in Lake Erhai and Xukou Bay of Lake Taihu compared with P. maackianus.

Furthermore, the ramet production of P. maackianus was inhibited more severely than V. natans by the decreased light regimes in the experiments, thus limiting the population expansion of P. maackianus in the low light, deep water environment. It has been reported that the rosette macrophyte V. natans has a decreased photosynthetic light compensation point and an increased C reservoir in low light environments10,22,24 and that canopy forming macrophytes usually have a relatively higher respiration rate than rosette producers25,26. This further facilitates V. natans abundance in low light and deep water environments. These results agree with our third hypothesis.

Additional Information

How to cite this article: Chen, J. et al. Differential photosynthetic and morphological adaptations to low light affect depth distribution of two submersed macrophytes in lakes. Sci. Rep. 6, 34028; doi: 10.1038/srep34028 (2016).