C3 plants
For C3 plants, the Acacia invasion did not induce changes in the Mussununga functioning. In non-invaded and invaded plots, the d13C increases as d15N increases and because high values of d13C are associated with high WUE (Werner and Máguas 2010), increasing WUE is also associated with increasing d15N. As hyperseasonal savannas have pulses of flooding and nitrogen is more available during the rainy season in the wet soil than in dry soil (see Chen et al. 1996), pulses of 15N-enriched nitrogen from higher trophic levels as termite mounds, dungs and urine deposition during rainy season possibly allow C3 plants to uptake a lot of nitrogen to grow with low nitrogen use efficiency (NUE) but with high WUE (high d13C) since WUE and NUE are negatively related (Field et al. 1983). Possibly, plants species with high d15N and d13C, especially small monocots plants with superficial roots, can grow in harsh sites of these hyperseasonal savannas getting and using N during pulses with low NUE but with high WUE as the soil surface dries.
In C3 plants, Leaf C% responds negatively to d13C and d15N, suggesting that plants with more structural C in leaves are not benefitting from N pulses during rainy season growing with high NUE and low WUE. Moreover, the plants with higher Leaf N% possibly do not use pulses of N availability during rainy season (low d15N) while maintaining high N-nutritional status (low C/N).
C4 plants
For C4 plants, the Mussununga functioning is not the same after invasion by Acacia. In non-invaded plots, C4 plants have d13C positively related to Leaf N% and negatively related to C/N. Therefore, the low d13C (low WUE) is a response to low Leaf N% and low N-nutritional status (high C/N) with an expected high NUE (Field et al. 1983). However, in invaded plots the d13C in C4 plants respond neither to Leaf N% nor to C/N indicating a loss of WUE response in an environment with N outflow from Acacia invaders (Meira-Neto et al. 2018b) that, possibly, enables C4 plants grow with low NUE. Congruently, the Leaf N% of C4 plants in invaded plots also lost the response to d13C suggesting that Leaf N% variation in invaded Mussunungas is related with N outflow into the invaded ecosystem (Rascher et al. 2012). Leaf C% also lost the response to d15N variation due to lack of response to d15N possibly because the invaded ecosystem is enriched with N with lower d15N from N-fixing invaders. The C/N (N-nutritional status) also lost responses to d13C and d15N, possibly influenced by N outflow from the Acacia invaders.
Hyperseasonal savanna functioning and consequences of Acacia invasion
In the current scenario of global change, not only the increasing concentration of CO2 and increasing temperatures can benefit C3 plants over C4 plants (Chen et al. 1996) in hyperseasonal savannas, but also the nitrogen-fixer invaders that cause loss of influence of N on d13C (WUE) in C4 plants with unpredictable consequences. Only two species are C4 in studied Mussunungas with most of the plants from only one sedge species. If the Acacia invasion is going to increase shading with increasing N contents in the Mussununga ecosystem (Meira-Neto et al. 2018b), the C4 plants will be possibly excluded from invaded Mussunungas since neotropical C4 plants do not tolerate shading (Klink and Joly 1989). However, the shading is not going to be caused by C3 herbs or monocots that are also shade intolerant. Instead, Acacia invaders and shade-tolerant C3 woody species are likely to shift hyperseasonal savannas into dense woodlands as the N outflow from Acacia invaders enriches the ecosystem with N (Meira-Neto et al. 2018b), especially if fire disturbance that boosts the invaders is frequent and the recruitment of Acacia invaders surpasses the recruitment of native species from hyperseasonal savannas (Meira-Neto et al. 2005; Le Maitre et al. 2011).
The plants with high d15N and d13C that use water efficiently and can grow in the harsh sites of Mussunungas often occur near termite mounds that are nitrogen sources (see Ji and Brune 2006) from a higher trophic level and with a higher d15N than leguminous N-fixers (Marshall et al. 2007).
There is a difference between C3 and C4, especially in invaded plots. In C3 plants, the higher d13C, the higher d15N and the lower Leaf C%. Also, the greater the need of C3 plants in N, the greater will be the d15N. Therefore, on the one hand the C3 plants that have low N-nutritional status depend on 15N-enriched sources of N to succeed in sites with strong N-limitation because of high WUE and high d13C, especially small monocots like Lagenocarpus rigidus and Actinocephalus ramosus. On the other hand, the C4 plants do not depend on 15N-enriched N, do not present significant relations between d13C (WUE) and d15N and, in invaded Mussunungas, the N-nutritional status (C/N) do not relate with d13C and with d15N.
The association between termite mounds and certain species with high d13C and high WUE in hyperseasonal savannas deserves attention because can shed light in an old controversy about the process in hyperseasonal savannas that originates Murundus, mounds attributed to termites or differential erosion, possibly with a facilitative influence of termite mounds according to our results (see Marimon et al. 2015). Moreover, apparently C3 plants that are 15N-enriched tolerate harsh sites that they would not tolerate without the N mineralization from termite mounds. But the origin of 15N-enriched N from higher trophic levels is not restricted to termites. Facilitative processes in African savannas associated to N-enriched dung of giraffes (high N:P ratio) or herbivores urine benefited grasses more than tree and shrub seedlings (Sitters and Venterink 2021) and termite mounds potentially cause similar effects on small monocots as Actinocephalus ramosus and Lagenocarpus rigidus especially because Mussunungas are used as pasture for extensive livestock in the region (Meira-Neto et al. 2005). Likely, the C3 monocots of the harshest sites of these hyperseasonal savannas, as Actinocephalus ramosus and Lagenocarpus rigidus, are dependent on 15N-enriched N from termite mounds, dungs and urine of herbivores (see Marshall et al. 2007; see also Sitters and Venterink 2021).
Actinocephalus ramosus presented the highest d18O possibly because of its lifeform, an annual plant, which could explain the high d18O value as its growth season coincides with the rainy season where the heaviest rains come from low altitude clouds enriched with 18O while rain of other seasons tends to be originated from higher clouds with lower d18O (see Marshall et al. 2007). The d18O profile of Blechnum serrulatum, a geophyte that resists without leaves in the soil during droughts and grows during rainy season, is similar to that of Actinocephalus ramosus. Marcetia taxifolia is a perennial shrub that presented intermediate values of d18O even though it can grow every time it rains. Gaylussacia brasiliensis and Acacia mangium have lower values of d18O as they are perennial and can also grow every time it rains. These results help to partially explain the remarkable growth of Acacia mangium in hyperseasonal savannas.
Mussununga, as a hyperseasonal savanna, does not accumulate water in the soil. Therefore, the entire volume of water is depleted by evapotranspiration or drainage in streams and does not drain water into aquifers available to plants. As a consequence, soil aquifers are not the cause of d18O variation in Mussununga plant species. Actinocephalus ramosus is an annual species, a C3 species with improved WUE as its d13C is higher (Werner and Máguas 2010) than most other C3 species but higher d13C can also be a result of a CAM facultative metabolism; however, d13C is not as high as the expected for CAM plants (Marshall et al. 2007). The high d18O of Actinocephalus ramosus could also be caused by the 18O-enriched rainwater that falls during its growing season, a d18O possibly boosted by a CAM facultative metabolism (Marshall et al. 2007). Actinocephalus ramosus can be a good indicator of hyperseasonal savannas as well as Marcetia taxifolia, another well-adapted species to hyperseasonal savanna that could also be CAM facultative because of intermediate values of d13C and high values of d18O.
Some C3 monocots as Lagenocarpus rigidus and Actinocephalus ramosus are 15N-enriched possibly because they are commonly associated to termite mounds, dungs and urine but they have low N-nutritional status, especially Lagenocarpus rigidus that is amongst the species with the highest C/N. Possibly, their high d13C means high WUE that depends on 15N-enriched N that is better mineralized and mostly available during pulses of N availability during rainy season for small C3 monocots. Blechnum serrulatum, a pteridophyte, is another species that benefits from the 15N-enriched in the same way of the small monocots Lagenocarpus rigidus and Actinocephalus ramosus.