Short-Term N-Fertilization Differently Affects the Leaf and Leaf Litter Chemistry of the Dominant Species in a Mediterranean Forest under Drought Conditions

: Nitrogen (N) deposition is a key driver of global change with signiﬁcant effects on carbon (C) cycling, species ﬁtness, and diversity; however, its effects on Mediterranean ecosystems are unclear. Here, we simulated N deposition in an N-fertilization experiment with 15 N-labeled fertilizer in a montane evergreen Mediterranean holm oak forest, in central Catalonia, to quantify short-term impacts on leaf, leaf litter elemental composition, and resorption efﬁciency in three dominant species ( Quercus ilex , Phillyrea latifolia , and Arbutus unedo ). We found that even under drought conditions, 15 N isotope analysis of leaf and leaf litter showed a rapid uptake of the added N, suggesting an N deﬁcient ecosystem. Species responses to N fertilization varied, where A. unedo was unaffected and the responses in P. latifolia and Q. ilex were similar, albeit with contrasting magnitude. P. latifolia beneﬁted the most from N fertilization under drought conditions of the experimental year. These differences in species response could indicate impacts on species ﬁtness, competition, and abundance under increased N loads in Mediterranean forest ecosystems. Further research is needed to disentangle interactions between long-term N deposition and the drought predicted under future climate scenarios in Mediterranean ecosystems.


Introduction
Nitrogen (N), which is an essential element for plant nutrition, is involved in complex cycling in terrestrial ecosystems that involves biotic and abiotic processes, such as N 2 -fixation, mineralization, nitrification, ammonification, and denitrification [1][2][3]. Anthropogenic activities since the industrial and agricultural revolutions, particularly those associated with fossil fuel combustion, cultivation of N 2 -fixing crops, and production of N fertilizer using the Haber-Bosch process, increasingly influence the global N-cycle as they are often key sources of N in terrestrial ecosystems, driving fluxes of similar magnitudes to natural fluxes [4][5][6]. Inorganic N deposition in European forests increased on average from about 2.8 kg ha −1 a −1 in 1900 to a peak of 10.3 Kg N ha −1 a −1 in 1990, followed by a decline to 6.6 Kg N ha −1 a −1 by 2018 [7,8]. However, 62% of Europe continues to remain at risk of eutrophication [9] through rises in organic N deposition due to increasing levels of N-fertilization of cropland [8].
In general, N addition to ecosystems alters the first N cycle, increasing N mineralization, nitrification, and nitrate leaching rates, and thereafter other ecosystem variables such as P-cycle [5], and finally is associated with a long-term plant community diversity decrease [10]. Also, in some Mediterranean regions in California with elevated N deposition, native plant species have been replaced by invasive species with higher The study was carried out on a south-facing slope (25% incline) of a semi-pristine (Quercus ilex L.) forest, where N deposition was 3.5 kg N ha −1 yr −1 between 1981 and 1994 [38], in the Prades Mountains, Catalonia, NE Spain (41 • 21 N, 1 • 02 E). The forest has not been disturbed for the last 70 years and the maximum height of the dominant species is about 6-10 m. Mean annual temperature and precipitation are 12 • C and 658 mm, respectively, with the warmest months occurring between June and August and the rainiest period between September and November ( Figure 1). These conditions correspond to a mesic-Mediterranean climate, with a pronounced summer drought period that usually lasts for 3 months. During the study period, mean annual temperature and annual precipitation in 2014 were 12.65 • C and 661.38 mm respectively, and 13.25 • C and 355.44 mm respectively in 2015 (Figure 1). Air temperature and rainfall were recorded by an automatic meteorological station in the studied area. The soil is a stony Dystric Cambisol, on a bedrock of metamorphic sandstone, and ranges between 35 and 100 cm in depth. The dense forest vegetation is dominated by Q. Ilex, with abundant Phillyrea latifolia L., Arbutus unedo L. among other evergreen species that are well-adapted to drought conditions (Erica arborea L., Juniperus oxycedrus L., and Cistus albidus L.) and the occasional occurrence of deciduous species (Sorbus torminalis L. Crantz and Acer monspessulanum L.). The three tree species (Q. ilex, P. latifolia, and A. unedo) represent 97% of the total aboveground biomass of the forest [39] and they frequently co-occur in Mediterranean maquis shrubland and evergreen Q. ilex forests. Q. ilex is widely distributed across the subhumid areas of the Mediterranean Basin, while P. latifolia occurs in warmer and drier Mediterranean areas [40,41] and A. unedo exploits forest gaps in which conditions are limiting for Q. ilex [42].
Forests 2021, 12, x FOR PEER REVIEW 3 of 17 forest vegetation is dominated by Q. Ilex, with abundant Phillyrea latifolia L., Arbutus unedo L. among other evergreen species that are well-adapted to drought conditions (Erica arborea L., Juniperus oxycedrus L., and Cistus albidus L.) and the occasional occurrence of deciduous species (Sorbus torminalis L. Crantz and Acer monspessulanum L.). The three tree species (Q. ilex, P. latifolia, and A. unedo) represent 97% of the total aboveground biomass of the forest [39] and they frequently co-occur in Mediterranean maquis shrubland and evergreen Q. ilex forests. Q. ilex is widely distributed across the subhumid areas of the Mediterranean Basin, while P. latifolia occurs in warmer and drier Mediterranean areas [40,41] and A. unedo exploits forest gaps in which conditions are limiting for Q. ilex [42].

Experimental Design
We established eight 15 m × 10 m plots at the same altitude (950 m) [43]; four plots received N-fertilizer and the remainder were untreated controls. In 2015, a total fertilization of 60 kg of N ha −1 a −1 (at a rate of 15 kg N ha −1 in each annual season) was applied as a solution of ammonium nitrate (NH4NO3) with a sprayer. This fertilization was sprayed in 3 different days per season (5 kg N ha −1 each application), with at least 1 week between different applications. This solution was sprayed over the canopy level after climbing to the top of the uppermost tree in each plot. This solution was enriched with 15 N as follows: 90% of "normal" ammonium nitrate, and 10% of ammonium nitrate enriched with 15 N (the 10% of this enriched ammonium nitrate was 15 N) (Sigma-Aldrich, Co., St. Louis, MO, USA).

Sampling
We sampled leaves of Q. ilex, P. latifolia, and A. unedo and foliar litter in mid-May (spring), end of July (summer), and mid-November (autumn) 2014, prior to treatment, and in 2015 during fertilization, sampling the same tree before and after treatment. On each sampling occasion, we randomly selected five trees of each species per plot and collected mature and healthy leaves at a height of 2-3 m from plants with a diameter of 2-12 cm at 5 cm above ground level; these plants represent about the 70% of the community biomass [44,45]. To reduce effects of tissue age and orientation to sunlight, we sampled leaves that were sun-lit and oriented southwards. Mean leaf life span is approximately 18 months [45], so most leaves sampled during 2015 under fertilization were also present during 2014 prior to treatment. Litterfall at the soil surface was collected in 20 circular baskets (27 cm in diameter, with 1.5-mm diameter mesh) that were randomly distributed in each plot. Even though the baskets were permanently placed over the ground, the dense canopy cover almost totally avoided N fertilization reaching the soil or the baskets. The collected litter was not in direct contact with the soil so its conservation state was good and the

Experimental Design
We established eight 15 m × 10 m plots at the same altitude (950 m) [43]; four plots received N-fertilizer and the remainder were untreated controls. In 2015, a total fertilization of 60 kg of N ha −1 a −1 (at a rate of 15 kg N ha −1 in each annual season) was applied as a solution of ammonium nitrate (NH 4 NO 3 ) with a sprayer. This fertilization was sprayed in 3 different days per season (5 kg N ha −1 each application), with at least 1 week between different applications. This solution was sprayed over the canopy level after climbing to the top of the uppermost tree in each plot. This solution was enriched with 15 N as follows: 90% of "normal" ammonium nitrate, and 10% of ammonium nitrate enriched with 15 N (the 10% of this enriched ammonium nitrate was 15 N) (Sigma-Aldrich, Co., St. Louis, MO, USA).

Sampling
We sampled leaves of Q. ilex, P. latifolia, and A. unedo and foliar litter in mid-May (spring), end of July (summer), and mid-November (autumn) 2014, prior to treatment, and in 2015 during fertilization, sampling the same tree before and after treatment. On each sampling occasion, we randomly selected five trees of each species per plot and collected mature and healthy leaves at a height of 2-3 m from plants with a diameter of 2-12 cm at 5 cm above ground level; these plants represent about the 70% of the community biomass [44,45]. To reduce effects of tissue age and orientation to sunlight, we sampled leaves that were sun-lit and oriented southwards. Mean leaf life span is approximately 18 months [45], so most leaves sampled during 2015 under fertilization were also present during 2014 prior to treatment. Litterfall at the soil surface was collected in 20 circular baskets (27 cm in diameter, with 1.5-mm diameter mesh) that were randomly distributed in each plot. Even though the baskets were permanently placed over the ground, the dense canopy cover almost totally avoided N fertilization reaching the soil or the baskets. The collected litter was not in direct contact with the soil so its conservation state was good and the decomposition process had hardly begun. Once collected, the leaf and litter samples were taken to the laboratory and stored at 4 • C prior to analysis.
Foliar nutrient resorption was calculated as the difference in elemental compound concentration of leaves collected in 2014 and litter collected in 2015, using the equation 100 − (X l × 100)/X f , where X l is litter element concentration in 2015 and X f foliar element concentration in 2014. The result is an inference of proportional (%) element resorption efficiency prior to leaf fall.

Chemical Analyses
The leaf and foliar litter samples were washed with distilled water, as described by Porter (1986) [46], dried in an oven at 60 • C to a constant weight, and then ground using a mill (CYCLOTEC 1093, Foss Tecator, Höganäs, Sweden) for measurement of biomass. Leaf and litter concentrations of carbon (C) and N were quantified by combustion coupled to gas chromatography, by placing 1.4 mg of the milled samples in a tin microcapsule in an elemental analyzer (CHNS-O EuroVector, Milan, Italy), while concentrations of phosphorus (P), potassium (K), and micro-elements were determined by digesting samples of leaf and litter in acid in a microwave (MARSXpress, CEM, Matthews, NC, USA) at high pressure and temperature. Then 250 mg of each ground sample, 5 mL of nitric acid, and 2 mL of H 2 O 2 were placed into Teflon tubes; digestate was placed in 50-mL flasks and diluted with Milli-Q water to a volume of 50 mL. Concentrations of P, K, and micro-elements were determined using optic emission spectrometry with inductively coupled plasma and the accuracy of the digestions and analytical biomass procedures was assessed using a certified standard biomass (NIST 1573a, tomato leaf; NIST, Gaitherburg, MD, USA) as a reference. Isotopic analyses of δ 13 C ( 13 C/ 12 C) and δ 15 N ( 15 N/ 14 N) of the leaf and litter samples were conducted at the Stable Isotope Facility at the University of California, Davis using a PDZ Europa ANCA-GSL elemental analyzer connected to a PDZ Europa 20−20 isotope ratio mass spectrometer (Sercon Ltd., Cheshire, UK).

Statistical Analyses
Between year treatment differences in plot, species, and season values were calculated based on proportional change in values between 2014 and 2015, and then linear mixed models were used to test for treatment differences in leaf and litter macro-(C, N, P, K) and micro-element concentrations (calcium, Ca; magnesium, Mg; chrome, Cr; iron, Fe; manganese, Mn; copper, Cu; zinc, Zn, strontium, Sr; nickel, Ni; and, lead, Pb) and reabsorption, and in foliar N and C isotopes by species, with plot and season as random factors. Models were defined using lme4 [47] and lmerTest [48] and r 2 was calculated in MuMIn [49] R packages. Treatment differences in leaf and litter isotope content by species was tested using Student's t-test.
Overall treatment differences in leaf and litter macro-(N, P, K) and micro-element (Ca, Mg, Fe, Mn, Cu, Zn) composition and isotope content (δ 13 C and δ 15 N) by species were tested using general discriminant analysis (GDA) in Statistica 8.0 (StatSoft, Inc., Tulsa, OK, USA) that identifies variables most responsible for group differences, while controlling variance due to other categorical variables (here, season).
In order to detect the effect of 2015's exceptional arid conditions, we calculated a t-test between 2015 and 2014 for leaves and litter from only control plots (which have no treatment applied in 2015).

Fertilization Effects on Foliar Element Concentrations
There were between-year differences in C:P ratios in P. latifolia, where there were increases of 4.83 ± 3.36% and 3.05 ± 0.86% in untreated control and fertilized plots, respectively (difference between treatments: p < 0.05; ±standard error (SE) showed; Figure 2a, Table S1), and ratios of N:P differed for A. unedo, where there were 1.64 ± 1.65% decreases and 1.12 ± 1.38% increases in the untreated and treated plots from 2014 (pretreatment) to 2015 (post-treatment), respectively, (difference between treatments: p < 0.05; ±standard error (SE) showed; Figure 1b, Table S1). There were no treatment differences in leaf concentration of N, P, or K or in C:N ratios of the three species.
Forests 2021, 12, x FOR PEER REVIEW 5 of 17 and 1.12 ± 1.38% increases in the untreated and treated plots from 2014 (pretreatment) to 2015 (post-treatment), respectively, (difference between treatments: p < 0.05; ±standard error (SE) showed; Figure 1b, Table S1). There were no treatment differences in leaf concentration of N, P, or K or in C:N ratios of the three species. Among the micro-elements, there were differences in leaf concentrations of chromium (Cr) in Q. ilex, where there were decreases of 1.68 ± 1.83% and 3.56 ± 0.68% in the untreated and treated plots, respectively (difference between treatments: p < 0.05; ±standard error (SE) showed; Figure 1c), while leaf concentration of nickel (Ni) in A. unedo decreased by 0.76 ± 0.41% and 1.49 ± 0.28%, respectively (difference between treatments: p < 0.05; ±standard error (SE) showed), and decreased in Q. ilex by 1.86 ± 1.79% and 4.67 ± 0.62% (difference between treatments: p < 0.05; ±standard error (SE) showed; Figure 2d) (Table S1). There were no treatment differences in leaf concentrations of Ca, Mn, Fe, Cu, Zn, Sr, or Pb in the three species.

Fertilization Effects on Leaf Litter Element Concentrations
Treatment affected leaf litter concentration of macro-elements only in P. latifolia (p < 0.05; ±standard error (SE) showed), where P concentration increased by 0.49 ± 0.59% in untreated plots and decreased by 1.80 ± 0.72% in fertilized plots (Figure 3a), while C:P ratios were decreased by 18.44 ± 3.21% and 7.12 ± 3.22% in untreated and fertilized plots, respectively (Figure 3b), and N:P ratios were 15.52 ± 3.85% and 4.11 ± 3.04% lower, respectively ( Figure 3c) (Table S2). There were no differences in leaf litter concentration of the remaining macro-elements in the three species.

Fertilization Effects on Leaf Litter Element Concentrations
Treatment affected leaf litter concentration of macro-elements only in P. latifolia (p < 0.05; ±standard error (SE) showed), where P concentration increased by 0.49 ± 0.59% in untreated plots and decreased by 1.80 ± 0.72% in fertilized plots (Figure 3a), while C:P ratios were decreased by 18.44 ± 3.21% and 7.12 ± 3.22% in untreated and fertilized plots, respectively (Figure 3b), and N:P ratios were 15.52 ± 3.85% and 4.11 ± 3.04% lower, respectively ( Figure 3c) (Table S2). There were no differences in leaf litter concentration of the remaining macro-elements in the three species.

Species Differences in Leaf and Leaf Litter Element Concentrations
The GDA driven by variation in δ 13 C, δ 15 N, K, Ca, Mn, Fe, Cu, and Zn concentrations (Table 1a) clearly separated the foliar elemental concentrations by species within and between treatments (Table 1b, Figure 6a). Similarly, there was clear separation of species in foliar litter within and between treatments (Table 2b, Figure 6b), driven by variation in δ 13 C, δ 15 N, N, P, K, Ca, Mn, and Fe concentrations (Table 2a). The variable loading most strongly in fertilized plots, separating them from control plots in leaves, is clearly δ 15 N and with lower intensity δ 13 C, while Cu, Mg, and Ca load to control plots. In litter, δ 15 N is the strongest variable loading in fertilized plots with N in lower intensity. Contrarily, Fe and Mn load to control plots.

Species Differences in Leaf and Leaf Litter Element Concentrations
The GDA driven by variation in δ 13 C, δ 15 N, K, Ca, Mn, Fe, Cu, and Zn concentrations (Table 1a) clearly separated the foliar elemental concentrations by species within and between treatments (Table 1b, Figure 6a). Similarly, there was clear separation of species in foliar litter within and between treatments (Table 2b, Figure 6b), driven by variation in δ 13 C, δ 15 N, N, P, K, Ca, Mn, and Fe concentrations (Table 2a). The variable loading most strongly in fertilized plots, separating them from control plots in leaves, is clearly δ 15 N and with lower intensity δ 13 C, while Cu, Mg, and Ca load to control plots. In litter, δ 15 N is the strongest variable loading in fertilized plots with N in lower intensity. Contrarily, Fe and Mn load to control plots.

Species Differences in Leaf and Leaf Litter Element Concentrations
The GDA driven by variation in δ 13 C, δ 15 N, K, Ca, Mn, Fe, Cu, and Zn concentrations (Table 1a) clearly separated the foliar elemental concentrations by species within and between treatments (Table 1b, Figure 6a). Similarly, there was clear separation of species in foliar litter within and between treatments (Table 2b, Figure 6b), driven by variation in δ 13 C, δ 15 N, N, P, K, Ca, Mn, and Fe concentrations (Table 2a). The variable loading most strongly in fertilized plots, separating them from control plots in leaves, is clearly δ 15 N and with lower intensity δ 13 C, while Cu, Mg, and Ca load to control plots. In litter, δ 15 N is the strongest variable loading in fertilized plots with N in lower intensity. Contrarily, Fe and Mn load to control plots.

Drought Effects
All species have differences in control plots elemental composition between 2014 and 2015 in foliar tissue and respond similarly to drought. In foliar tissue, all macro-nutrients decrease significantly in 2015 in all three studied species except C in P. latifolia, which have no significant differences. In foliar micro-nutrients, Mg and Cu decrease significantly in 2015 in all three species, Ca increases significantly in P. latifolia and Q. ilex; Cr and Ni decreases in A. unedo and Q. ilex; Mn and Sr increase significantly in all three species; Fe decreases in A. unedo, increases in Q. ilex and has no differences in P. latifolia; Zn decreases in Q. ilex; and Pb decreases in A. unedo and P. latifolia and has no differences in Q. ilex (Figure 7).

Drought Effects
All species have differences in control plots elemental composition between 2014 and 2015 in foliar tissue and respond similarly to drought. In foliar tissue, all macro-nutrients decrease significantly in 2015 in all three studied species except C in P. latifolia, which have no significant differences. In foliar micro-nutrients, Mg and Cu decrease significantly in 2015 in all three species, Ca increases significantly in P. latifolia and Q. ilex; Cr and Ni decreases in A. unedo and Q. ilex; Mn and Sr increase significantly in all three species; Fe decreases in A. unedo, increases in Q. ilex and has no differences in P. latifolia; Zn decreases   In foliar litter elemental composition, the differences between 2014 and 2015 likely due to drought vary more between species than in foliar elemental composition. A. unedo has significant decreases in P, C:N, Mg, Ca, Zn, and Sr, and significant increase in Ni. P. latifolia has significant increases in N, Cr, Mn, Fe, and Pb, and significant decreases in C, In foliar litter elemental composition, the differences between 2014 and 2015 likely due to drought vary more between species than in foliar elemental composition. A. unedo has significant decreases in P, C:N, Mg, Ca, Zn, and Sr, and significant increase in Ni. P. latifolia has significant increases in N, Cr, Mn, Fe, and Pb, and significant decreases in C, K, C:N, C:P, N:P, Mg, Ca, Cu, Zn and Sr. In Q. ilex, there is a significant increase only in Fe and significant decreases in K, C:N, C:P, N:P, Mg, Ca, Ni, and Zn (Figure 7).

Interannual Differences
The overall macro-elements reduction and micro-elements variation in control plots between 2014 and 2015, as seen in Figure 7, are inherent effects of the exceptional drought in 2015. These year to year effects of drought are consistent with the drought experiments carried out in the same studied Mediterranean forest, where reduction of foliar N [50,51], P [52][53][54], and micro nutrients variation [55,56] were described to be associated to the drought treatment. This decrease of nutrient uptake associated to a decrease of soil water content gives evidence of the potential negative effects of drought by decreasing nutrient uptake capacity as well as the reinforced negative feedback of low nutrient concentration in leaves lowering the plant capacity to maintain an adequate water use efficiency [57]. Therefore, to correctly assess the effects of addition of N in this study, the exceptional 2015 arid conditions needs to be considered.

Macro-Elements
In contrast to our results, fertilization with N would be expected to increase total concentration of foliar N; however, we found evidence of an increase in foliar 15 N that indicates rapid plant uptake of the applied fertilizer. The contrasting effects of added N on total foliar N concentration and foliar 15 N may be explained by two hypotheses. First, added N is highly available, so although plants in the fertilized plots may have taken up similar amounts of total N from soil, there may have been greater uptake of the more easily available rich 15 N, and secondly, plants may have taken up greater amounts of N that was rapidly allocated to tissue other than leaf material, such as roots, to stimulate growth. Our results support these hypotheses, because we observed higher foliar concentrations of other elements following fertilization and higher leaf litter 15 N concentrations that indicate decreases in leaf N resorption under increased N-availability.

Micro-Elements
In previous studies, N addition was related with a massive incorporation of NO 3 − into the system, causing cation losses by leaching and soil acidification [58][59][60]. In our study, leaching could have been generated by direct NO 3 − contact to the leaves, as described for acid rain impacts [61]. The decrease of foliar Cr in Q. ilex (Figure 2d) and Ni in A.unedo and Q. ilex (Figure 2e) is consistent with foliar leaching theory. In contrast, litter micro-elemental concentrations were not conclusively responding to N addition. The most remarkable response of micro-elements to N addition was the decrease of foliar Cr and Ni, which do not have a known biological benefit to plants and are considered pollutants. In this case, the short-term effects of N addition may be considered beneficial for species fitness, even though the size of effect varied among the three species (greatest in Q. ilex, moderate in A. unedo, and null in P. latifolia).

Nutrient Uptake
The isotope analysis showed rapid incorporation of N to the ecosystem following fertilization, where there were differences in foliar δ 15 N, but not of δ 13 C. Foliar uptake of N by A. unedo was unaffected by treatment, whereas fertilization led to greater uptake in P. latifolia and Q. ilex (Figure 6), indicating possible interspecific imbalances in nutrient uptake under drought conditions. In contrast, leaf litter 15 N was consistently greater among the three species under N-fertilization, where the greater concentration in A. unedo and Q. ilex leaf litter than in leaf material highlights differences in resorption and indicates that, prior to the release of senescent leaves, A. unedo and Q. ilex reabsorbed smaller or negligible amounts of the applied N enriched with 15 N. There were no differences between δ 15 N of P. latifolia leaf material and leaf litter, indicating unchanged levels of N resorption under N addition; this maintenance of greater nutrient use efficiency, even under drought conditions, than in A. unedo and Q. ilex, is consistent with previous studies that report its greater capacity for the maintenance of optimal function and growth under increasing drought than the other two species [44,62].

N limited Ecosystem
Our analyses showed rapid leaf and leaf litter elemental composition responses to N addition, through the increases in leaf and leaf litter δ 15 N, allowing us to conclude that this Mediterranean forest ecosystem is N limited, which is consistent with previous studies that showed decreases in foliar N in Mediterranean ecosystems during the 20th century, and associated them to increases in photosynthetic rates and C fixation, as well as growth capacity because of increasing atmospheric CO 2 concentrations [63] and with N limitation reported in the forests we studied [64]. Furthermore, the drought in 2015 also contributed to decreasing foliar N and increased even more the original plant N deficiency phenomenon already described in other studies [65].
These three dominant species of the holm oak forest ecosystem were characterized by smaller decreases in leaf and leaf litter concentrations of several elements under Nfertilization during the extreme high drought conditions of 2015. This rapid response of the mature ecosystem was remarkable and indicates that addition of N to an N-limited ecosystem improves nutrient retention and increases resilience of ecosystem function to drought stress. Such drought recovery under N fertilization conditions could drive forest managers to consider generalized application of N-fertilization to stimulate C-uptake after droughts. However, N fertilization would be logistically complicated and would cause important N-cycle alterations, nutrient imbalances, decrease of diversity, and also can favor highly competitive invasive species replacing endemic ones. However, these impacts could be improved adding other nutrients such as P and K or even Fe, Mg, and sulfur. In this way, the fertilization probably would not generate important nutrient imbalances, and plants with better nutritional status could improve their water use efficiency. Our results can thus be useful for developing management strategies for Mediterranean forests under increasingly dry conditions projected for coming decades.
Long-term continuous N-fertilization experiments are needed to confirm the wider effects of increased availability of N on growth and stocks and changes in plant-soil cycling of nutrient elements of this Mediterranean ecosystem, including its dominant species. It is likely that sustained N deposition would decrease N limitation and its associated effects, but elements that derive from the bedrock, such as P or K, will increasingly become leached and limiting [5,6].

Biogeochemical Niche Differentiation
The GDA analysis showed a clear differentiation in element concentration among the three studied species and with treatment. This differentiation suggests a particular chemical identity per species as well as a particular species response to the same drivers. This chemical differentiation among species is supported by biogeochemical niche hypothesis [33,66,67]. Biogeochemical niche hypothesis is a derivation of the classic niche theory, which assigns to each species a specific position into a multidimensional space of traits, environmental requirements, and "needs". Biogeochemical niche hypothesis aims to synthesize all these variables into an easily quantifiable multidimensional space of the elemental composition with the "n-dimensional" axes being the concentrations of the different bioelements. This chemical divergence responds to the necessity of avoiding niche overlap that has been driven through evolutive genetic selection and the environmental conditions where the species has been developed.

Species Differences in Response to N Fertilization
A. unedo, P. latifolia, and Q. ilex coexist at the same stage of ecological succession in the Mediterranean forest ecosystem [42,68], presenting different intraspecific tolerance to different extreme conditions. Despite P. latifolia and Q. ilex sharing the same strategy to avoid water loss during drought, P. latifolia has a higher tolerance, a more efficient recovery to drought [69], and high temperatures than Q. ilex [44]. At the same time, P. latifolia has lower tolerance to low winter temperatures than Q. ilex [69]. Also, previous studies of impacts of long-term drought have shown that Q. ilex and A. unedo are mostly affected through changes in C and nutrient storage, whereas P. latifolia tends to remain unaffected [44,62]. Despite all three species sharing a dimorphic root system, enabling them to access different water level sources [70], these differences seem to be explained through xylem vessels width. Q. ilex and A. unedo have wider xylem vessels, which allow them to transport greater amounts of sap under well-watered conditions, but makes them more susceptible to embolism under drought conditions [71]. Also, Q. ilex has higher capacity to exclude potential toxic ions than A. unedo [72].
The intraspecific different responses to N-fertilization observed in this study are consequent with the different species tolerance to drought conditions, where P. latifolia is the most drought-resistant species and also the one best at assimilating better N addition. As long as the fertilizer is sprayed to the leaves, its absorption is directly related to stomatal opening. Thus, better watered species can afford longer stomatal apertures and consequently, higher N absorption [73]. This different N absorption among species may lead to changes in species composition in a drier and N fertilized environment as projected for these Mediterranean ecosystems for the next decades [9,28,74]. Under extreme drought conditions, P. latifolia could overtake Q. ilex dominance [65].

Conclusions
We found that added N was rapidly utilized in the three studied species, consistent with previous observations of N limitation in this area; this effect was remarkable, given the exceptional arid conditions during the study. Species responses to sudden increases in N availability varied, where A. unedo was unaffected and direction of responses in P. latifolia and Q. ilex were similar, albeit with contrasting magnitude. The macro-element concentrations were more sensitive in P. latifolia than in the other two species. P. latifolia was the species that took best advantage of the N fertilization and the species less affected by drought. The chemical composition of the three study species before and after Nfertilization follows the biogeochemical niche hypothesis, where differences in response may trigger changes in species coexistence and community composition.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/f12050605/s1, Table S1: Results of the foliar mixed models. Table S2: Results of the litter mixed models. Table S3: Results of the reabsorption mixed models. Table S4: Results of the isotopes mixed models.