The effect of nitrogen availability and water conditions on competition between a facultative CAM plant and an invasive grass

Abstract Abstract Plants with crassulacean acid metabolism (CAM) are increasing their abundance in drylands worldwide. The drivers and mechanisms underlying the increased dominance of CAM plants and CAM expression (i.e., nocturnal carboxylation) in facultative CAM plants, however, remain poorly understood. We investigated how nutrient and water availability affected competition between Mesembryanthemum crystallinum (a model facultative CAM species) and the invasive C3 grass Bromus mollis that co‐occur in California's coastal grasslands. Specifically we investigated the extent to which water stress, nutrients, and competition affect nocturnal carboxylation in M. crystallinum. High nutrient and low water conditions favored M. crystallinum over B. mollis, in contrast to high water conditions. While low water conditions induced nocturnal carboxylation in 9‐week‐old individuals of M. crystallinum, in these low water treatments, a 66% reduction in nutrient applied over the entire experiment did not further enhance nocturnal carboxylation. In high water conditions M. crystallinum both alone and in association with B. mollis did not perform nocturnal carboxylation, regardless of the nutrient levels. Thus, nocturnal carboxylation in M. crystallinum was restricted by strong competition with B. mollis in high water conditions. This study provides empirical evidence of the competitive advantage of facultative CAM plants over grasses in drought conditions and of the restricted ability of M. crystallinum to use their photosynthetic plasticity (i.e., ability to switch to CAM behavior) to compete with grasses in well‐watered conditions. We suggest that a high drought tolerance could explain the increased dominance of facultative CAM plants in a future environment with increased drought and nitrogen deposition, while the potential of facultative CAM plants such as M. crystallinum to expand to wet environments is expected to be limited.


| INTRODUCTION
Climate change studies predict an intensification of drought in many drylands around the world (Easterling et al., 2000;IPCC 2013). Human activities associated with fertilizer applications have dramatically increased atmospheric nitrogen deposition, a trend that is expected to continue in the decades to come (Goulding et al., 1998;Galloway et al., 2008). These key global change drivers (i.e., altered water and nutrient conditions) have been found to greatly affect ecological processes such as interspecific interactions in ecosystems where C 3 and/or C 4 plants dominate (Niu, Liu, & Wan, 2008;Van der Waal et al., 2009). It remains unclear, however, how water and nutrient conditions affect plants with crassulacean acid metabolism (CAM) and their competitive relationship with other functional types (but see Yu & D'Odorico, 2015Yu, D'Odorico, Li, & He, 2017).
This study investigated the effects of nutrient and water conditions on competition between the model facultative CAM species Mesembryanthemum crystallinum and its C 3 competitor Bromus mollis, which co-occur in California's coastal grasslands (Vernon, Ostrem, Schmitt, & Bohnert, 1988;Schmitt, 1990;Winter & Holtum, 2014).
M. crystallinum is native to southern and eastern Africa and was introduced to, and then spread throughout, western Australia, the Mediterranean basin, and along the coasts of the western United States, Mexico, and the Caribbean (Adams et al., 1998). In natural habitats, it germinates and establishes in the rainy season with C 3 photosynthesis and then switches to CAM photosynthesis in response to environmental stress (e.g., low water and/or high salinity). This switch in photosynthetic pathway occurs when M. crystallinum transitions from juvenile to adult (≈>6-7 weeks old) and develops secondary/ succulent leaves alongside shoots (Osmond, 1978;Winter & Holtum, 2007, 2014. This CAM behavior in M. crystallinum can revert back to C 3 photosynthesis after removing the source of environmental stress, thus demonstrating the crucial role of environmental controls in CAM behavior (Vernon et al., 1988;Schmitt, 1990;Winter & Holtum, 2014).
Field observations show that M. crystallinum appears to be increasing its abundance in coastal California (Vivrette & Muller, 1977;Corbin & D'Antonio, 2009), while its potential expansion in a future changing environment (i.e., increased drought and nitrogen deposition) remains unclear.
Mesembryanthemum crystallinum with high drought tolerance would be expected to outcompete B. mollis in a future drier climate.
Increased N deposition would increase high growth rates and water usage of grasses (B. mollis) in the wet (rainy) season (McCown & Williams, 1968), which increases water stress of grasses in the subsequent dry season, and thus potentially favors M. crystallinum by releasing competitive pressure from grasses. In comparison, in consistently wet conditions B. mollis could sustain higher growth rates, especially in high N conditions, and thus have a competitive advantage with respect to M. crystallinum in access to light, soil water, and nutrients. This competitive advantage by B. mollis may exert biotic stress on M. crystallinum. However, it remains unclear whether M. crystallinum may adapt to biotic stress from B. mollis by switching to CAM photosynthesis, a strategy of photosynthetic plasticity found to increase its reproduction rate and fitness (Winter & Ziegler 1992;Cushman et al., 2008;Herrera, 2009). Testing these novel hypotheses would provide new insights into crassulacean acid metabolism as an adaptive strategy to both abiotic and biotic stress. The ability of facultative CAM plants to adapt to strong competition enhances their potential to expand in wet environments, an aspect that has been largely ignored in the past decade.
Another knowledge gap is the effects of nutrient availability (mainly N) on CAM expression in M. crystallinum. Some studies in obligate CAM species indicate that N deficiency limited the rate of CAM photosynthesis (Winter, Foster, Schmitt, & Edwards, 1982;Nobel, 1983), presumably because of the N requirements by the enzymes used for photosynthesis. In contrast, other studies show that CAM plants (including both obligate and facultative) grown in conditions with lower N availability had a higher CAM expression (Ota, 1988;Paul & Cockburn, 1990;Lüttge, 2006;Winter & Holtum, 2011).
Recent studies recognized the role of carbohydrates (i.e., 3-carbon acceptor phosphoenolpyruvate, PEP which is produced by degrading starch/sugars) as substrates in nocturnal carboxylation (Borland & Dodd, 2002;. Haider, Barnes, Cushman, and Borland (2012) found that CAM expression in a starch-deficient mutant of M. crystallinum was suppressed under high salt additions. Indeed, the only study to investigate the effects of N and P deficiency on CAM expression in M. crystallinum found that CAM expression increased (Paul & Cockburn, 1990). However, the N and P deficiency was applied at the adult stage, after the plants were able to accumulate a sufficient amount of carbohydrates from earlier growth stages with no N and P limitation. It is unclear how nutrient treatments applied in early life stages and their interactions with water availability could affect CAM expression in M. crystallinum.
We conducted greenhouse experiments in which the seedlings of M. crystallinum and B. mollis in both monoculture and mixtures were subjected to two nutrient levels ("high" and "low") and two water levels ("high" and "low"). Plant responses were evaluated through measurements of gas exchange, concentrations of titratable acidity, aboveground plant N, biomass, and productivity. We asked: (i) How does competition between M. crystallinum and B. mollis respond to nutrient and water conditions? (ii) Is the physiological plasticity of facultative CAM in M. crystallinum an adaptive strategy for competition with B. mollis? (iii) How does nutrient availability and its interactions with water availability influence CAM expression and reversibility in M. crystallinum? The study used a randomized block experiment design in which the seedlings of M. crystallinum and B. mollis in both monoculture and mixture were subjected to two nutrient fertilization levels (high and low) under high and low water conditions. This experiment had three stages of plant harvest (Figure 1), and there were six replicates arranged in six blocks for each measurement in each stage. Fertilizer was applied in the form of Peters Professional 20-20-20 (20% total N including 3.2% NH 4 -N, 5.3% NO 3 -N, and 11.5% urea, 20% P 2 O 5 , 20% K 2 O, as well as other micronutrients). Each pot in high nutrient conditions received 15 mg N once every 8 days while each pot in low nutrient conditions received 15 mg N once every 24 days. Plants in high water conditions were watered every 2 days with an intensity of 8 mm per event during the whole experiment. Plants in low water conditions were watered every 2 days with an intensity of 8 mm per event until 19 October 2015 and were then watered every 8 days with an intensity of 8 mm per event until 6 December 2015, when the water treatment returned to high water conditions (i.e., watering frequency of once every 2 days) until the end of the experiments (5 January 2015); thus, this low water treatment was in the form of a "wet-drywet" sequence ( Figure 1). We note that, in contrast to previous studies (Winter & Holtum, 2007, 2014, the low water treatment entailed a low watering frequency instead of complete interruption of water applications. The competitive advantage of M. crystallinum over B. mollis and its successful invasion of coastal California has been attributed to its high salt tolerance (Vivrette & Muller, 1977). Over this experiment salts were not added to increase soil salinity because the purpose of this study was to investigate the potential expansion of M. crystallinum under increased drought and nitrogen deposition. Plants were subjected to well controlled light (i.e., 12-hr light with photosynthetically photon flux density (PPFD) of 700-800 μmol m −2 s −1 from 7 a.m. to 7 p.m. EDT) and temperature (i.e., 25°C during the day and 20°C during the night) conditions.

| Gas exchange and titratable acidity measurement
Over the experiment, M. crystallinum produces regular/secondary leaves, which allows the measurements of gas exchange using a standard leaf chamber (2 × 3 cm 2 ) in a Licor 6400 gas analyzer.
Before plant harvest (usually 2 days before), gas exchange for leaves of M. crystallinum and B. mollis in both monoculture and mixture were measured using the standard Licor leaf chamber. During the day, the measurements of gas exchange were made for M. crystallinum and B. mollis between 10:00 a.m. and 12:00 p.m. at a constant leaf temperature of 23°C and photosynthetically active radiation of 1,500 μmol m −2 s −1 ; during the night, gas exchange was measured once every 2 hr between 8 p.m. and 8 a.m. on the following day for M. crystallinum-both alone and mixed with B. mollis-at a constant leaf temperature of 23°C and photosynthetic active radiation of 0 μmol m −2 s −1 . Thus, gas exchange was measured once during the day and several times at night to capture the nocturnal change of CO 2 uptake.
After measurements of gas exchange (still before plant harvest), in each treatment six leaves from six individuals of M. crystallinum (alone or mixed with B. mollis; one leaf each individual) were sampled from each plant at 7 a.m. and 5 p.m., respectively, and then stored at −20°C before measurements of titratable acidity. CAM photosynthesis is characterized by a temporal separation of the dark (i.e., accumulation of 4-C organic acids using phosphoenolpyruvate carboxylase to fix CO 2 ) and light reactions of photosynthesis (i.e., decarboxylation of 4-C organic acids for Calvin cycle using Rubisco). Thus, a significant increase in titratable acidity overnight indicates the occurrence of CAM photosynthesis in M. crystallinum. Titratable acidity was measured using the acid base titration method (Von Caemmerer & Griffiths, 2009), whereby leaf disks (4 cm 2 ) are boiled in 1.5 ml H 2 O for 5 min in a microfuge tube; 10 mmol/L NaOH was added into the same tube with 20 μl of a 1/5 dilution of phenolphthalein as indicator. Titratable F I G U R E 1 Schematic diagram of water treatments in high and low water conditions. Black zone represents low-frequency watering treatment (once every 8 days) while the white zone represents high-frequency watering treatment (once every 2 days). Plants were harvested in the first (October 30-31th), second (December 4-5th), and third (January 4-5th) stages of the experiment. Gas exchange and titratable acidity were measured 1-2 days before each harvest

| Light availability and biomass measurements
Before plant harvest, light intensity in mixture was measured above and at the bottom of canopies (approximately at ground level) using a HOBO Pendant® Temperature/Light 64K Data Logger. Relative light intensity (%) was calculated as the ratio of light intensity under canopies to that above canopies (Sun et al. 2016). Plants were harvested on Total biomass and shoot-to-root biomass ratios were calculated. Note that the samples of fresh M. crystallinum collected for measurements of titratable acidity were weighted and then converted to dry biomass using the fresh/dry biomass ratio, based on our measurements.

| Plant leaf water potential and plant N content
Plant leaf water potential was measured using a Decagon WP4® potentiometer. Plant samples dried at 60°C for 72 hr were ground and homogenized for elemental analysis. Plant N analysis was performed using a Thermo Scientific FLASH 2000 NC Analyzer.

| Statistical analysis
The effects of nutrient treatment, water availability, species competition and time, as well as their interactions, on plant leaf water potential (LWP), specific leaf area (SLA), diurnal photosynthetic assimilation (A D ), total biomass (TB), belowground-to-aboveground biomass ratio (BA), and aboveground plant total N (APN) were analyzed using a fiveway ANOVA with block as a random factor. The effects of nutrient, water, species, and time as well as their interactions on soil moisture were analyzed using a four-way ANOVA with block as a random factor. The BA values were natural log transformed prior to ANOVA. In general, the most interesting effects were found in multiway interactions. To explore these interactions, we constructed pairwise orthogonal contrasts to detect differences between individual pairs of means.
All statistics were performed in SAS 9.4.

| Competition between Mesembryanthemum crystallinum and Bromus mollis
Leaf water potential in M. crystallinum alone (FC) was significantly greater than in B. mollis alone (G) and B. mollis in mixture (GM), HNHW, high nutrient and high water conditions; LNHW, low nutrient and high water conditions; HNLW, high nutrient and low water conditions; LNLW, low nutrient and low water conditions. Each bar represents the mean of six values while error bars indicate 95% confidence intervals especially in low water conditions over all stages of the experiment (all p ≤ 0.0158, Figure 2). B. mollis both alone and in mixture died at some time between the first and second stage, as shown by its extremely low leaf water potential (Figure 2a,b), lack of photosynthetic assimilation (Figure 3a,

| Effects of nutrient availability, water conditions, and competition on plant response
Leaf water potential (Figure 2), photosynthetic assimilation There was a significant effect of nutrient and water interactions in affecting leaf water potential, photosynthetic assimilation, and total biomass (Table 1; p < 0.001 for N × water). Similar to the pattern of specific leaf area (Fig. S3), in high water conditions, A D of all vegetation types in high nutrient conditions was significantly greater than low nutrient conditions (all p ≤ 0.0387, Figure 3). In low water

| Competition between Mesembryanthemum crystallinum and Bromus mollis
We found that M. crystallinum outcompeted B. mollis in low water treatments regardless of nutrient availability (Figures 2-4). The death of B. mollis in low water treatments was mainly caused by its intolerance to water stress exacerbated by the relatively high density (20 individuals per pot) and high biomass accumulated during prior high water conditions, which led to high evapotranspiration.
As compared to B. mollis, M. crystallinum had much higher leaf water potential ( Figure 2) and was more tolerant to water stress likely because of its ability to store the absorbed water in aboveground biomass (Lüttge, 2004;Borland et al., 2009  increase. In addition to drought intensification, another driver of environmental change is increased N deposition. Previous studies suggested that Mediterranean ecosystems such as California grasslands where M. crystallinum and B. mollis interact could be particularly vulnerable to impacts from climate change and N deposition (Parton, Ojima, & Schimel, 1994;Sala, 2000). This study has shown that increased nutrient availability and low water availability affected the competitive relationship between M. crystallinum and B. mollis. In fact, we found that after ceasing water applications (stage 1 of the experiment) the T A B L E 1 Results (p values) of five-way factorial ANOVA on total biomass (TB), belowground-to-aboveground biomass ratio (BA), photosynthetic assimilation during the day (A D ), plant leaf water potential (LWP), specific leaf area (SLA), and aboveground plant N (APN) leaf water potential of B. mollis-both alone and in mixture-was much lower with high rates of nutrient supply than in low nutrient conditions ( Figure 3a); we also observed that B. mollis in high nutrient conditions died earlier (≈1-2 weeks) than in low nutrient conditions in response to drought treatments. Relatively high levels of nutrient availability increased the biomass of B. mollis, and consequently led to higher evapotranspiration rates and associated soil moisture depletion, thereby enhancing plant water stress after watering frequency was reduced, consistent with other studies (Zavaleta et al., 2003;Harpole, Potts, & Suding, 2007).
Nutrient level and water co-limited the photosynthesis and productivity of M. crystallinum and B. mollis (Figures 3 and 4). These results were consistent with various studies of grasslands across a large range of precipitation regimes (Harpole et al., 2007;Eskelinen & Susan 2015). Moreover, we found much lower leaf water potential (Figure 2), aboveground plant N (Fig. S1), light availability (Fig. S2), photosynthetic assimilation, and total biomass (Figures 3 and 4) in M. crystallinum in mixture than alone; thus, B. mollis exerted a strong competition effect on M. crystallinum for access to soil nutrients and light in high water conditions. This competition effect even outweighed the positive direct effects of increased nutrient availability on photosynthetic assimilation and total biomass of M. crystallinum (Figures 3 and 4) and the reduction in root/shoot ratio (Fig. S4). In fact, high-stature B. mollis took advantage of increased nutrient availability and constrained the growth of low-stature M. crystallinum by enhancement of shade effects (Yang et al., 2011;Sun, Yu, Shugart, & Wang, 2015). In agreement with other studies (Tilman, 1988;Lane, Coffin, & Lauenroth, 2000;Harpole et al., 2007), M. crystallinum in response to light competition increased the biomass allocation to aboveground (Fig. S4), suggesting a shift in limiting resources from belowground (nutrients) to aboveground (light).

| CAM expression and reversibility in Mesembryanthemum crystallinum as affected by competition
Surprisingly, in high water conditions, M. crystallinum in mixture did not switch from C 3 photosynthesis to CAM expression over the entire experiment, a type of physiological plasticity M. crystallinum typically uses to adapt to environmental stress (Osmond, 1978;Winter & Holtum, 2007, 2014 and increase production of seeds and overall fitness (Cushman et al., 2008;Herrera, 2009 (Winter & Holtum, 2007, 2014. Alternatively, in high water conditions photosynthesis and productivity of M. crystallinum in association with B. mollis was substantially suppressed (Figures 3 and 4) and thus did not have sufficient carbohydrates reserves

| CAM expression and reversibility in Mesembryanthemum crystallinum as affected by abiotic factors
Our research shows that CAM expression in M. crystallinum was the result of environmental controls in response to droughts, consistent with other studies (Vernon et al., 1988;Schmitt, 1990;Piepenbrock & Schmitt, 1991;Winter & Holtum, 2007, 2014. This was demonstrated both by the lack of CAM expression in M. crystallinum during the whole three stages in high water conditions, and the reversibility of CAM expression after returning to high water conditions in low water treatments. Previous studies investigated the effects of N deficiency and its interactions with light on CAM expression in CAM plants, but they reported a mixed response: in some cases, N deficiencies had a negative and in others a positive effect on CAM photosynthesis (Winter et al., 1982;Nobel, 1983;Ota, 1988). Our study shows that in low water conditions, there was no significant difference in CAM expression by M. crystallinum between high nutrient and low nutrient conditions ( Figure 5). Paul and Cockburn (1990) applying adequate N and P supply to young seedlings, leading to sufficient photosynthesis and carbohydrate accumulation, found a positive response of CAM expression in adults of M. crystallinum to N and P deficiency. In our study, low nutrient availability (one-third of high nutrient conditions) was applied during all stages; in low water conditions, photosynthesis and production of carbohydrate are mainly limited by soil moisture instead of nutrients (Figures 3 and 4), implying that low nutrient supply is still adequate relative to water conditions and could sustain sufficient carbohydrate for nocturnal carboxylation.
Although low nutrient supply was a limiting factor of plant photosynthesis and productivity in high water conditions (Figures 3 and 4), CAM expression in M. crystallinum did not occur because of lack of water stress ( Figure 5; Osmond, 1978;Winter & Holtum, 2007, 2014. These results were in line with the studies that showed that CAM was best expressed in facultative Kalanchoe lateritia at moderately low N conditions (i.e., with N applications 20% of the reference rate) as compared to ambient and very low N availability, likely because sufficient carbohydrate and environmental stress (i.e., N deficiency) were both satisfied under moderately low N (Santos & Salema 1991. This study investigated the ecophysiological mechanisms underlying the potential advantage of a model facultative CAM species (M. crystallinum) while interacting with a C 3 species (B. mollis) found in California's coastal grasslands under nutrient and water manipulations. We found that because of its drought tolerance, M. crystallinum had a competitive advantage relative to B. mollis in low water and N deposition conditions. In high water conditions, however, B. mollis was a stronger competitor for soil nutrients and light resources. This strong competition restricted the ability of M. crystallinum to switch to CAM expression-a type of physiological plasticity used by M. crystallinum to adapt to environmental stress and to increase seed production and plant fitness. With an increasing trend of drought and nitrogen deposition in the coming decades, we suggest that the dominance of CAM plants is likely to increase in drylands in a future environment. The lack of ability to adapt to strong competition by switching to CAM photosynthesis in high water conditions suggests that the potential of facultative CAM plants such as M. crystallinum to expand to wet environments would be restricted.

This research was funded through a fellowship from China Scholarship
Council, a grant from the VPR Office of the University of Virginia, and a grant from the National Socio-Environmental Synthesis Center, NSF DBI-1052875. We would like to thank Wei Li, Yongli He, and Wendy Crannage for their help in experiments.

AUTHOR CONTRIBUTIONS
K.Y., P.D., and S.C. designed the research; K.Y. developed and performed experiments; K.Y., D.C., and A.P. analyzed data; K.Y. wrote the manuscript; and all others contributed to revisions.