Structural responses to salt accumulation in foliage
As indicated by significant correlation with both Na (P <0.001, r=0.79) and Cl (P =0.0007, r=0.67), the leaves of lime trees from Riga’s street greeneries showed a higher percentage area in the form of necrotic leaf rims (Fig. 2) with increasing salt accumulation. However, the LMEM models were not significant (Table S1B). In parts of leaves still asymptomatic, structural changes clearly distinct from those by biotic injury – noteworthy aphids (Bouraoui et al., 2019) - were observed in the leaf mesophyll (Fig. 6) and epidermis (Fig. S2). The most striking symptom of salt accumulation was an increase in the percentage area of vacuome (Fig. 6D-F), with one large vacuole filling most of cell volume in samples with highest contamination (Fig. 6I, J). Given the ontological progress in epidermis by the time of sampling and random sectioning of spongy parenchyma cells, this marker was primarily visible in palisade parenchyma. Other palisade cell structures were increasingly condensed and degenerated (Fig. 6G-J, L, M). At subcellular level, the vacuoles showed intense autophagic activity and contained many inclusions as well as multivesicular bodies (Fig. 6I, J, L, M). Large plastoglobules, protruding towards and being extruded into the vacuole (Fig. 6L, M), had developed inside of chloroplasts. In the epidermis, mostly degenerative changes including the a) higher condensation of nuclear material (Fig. S2B, E), b) increase in size and frequency of plastoglobules within leucoplasts (Fig. S2F, G) or c) larger autophagic activity and frequency of multivesicular bodies were observed.
The LMEM models in combination with salt microlocalisation data confirmed the direct or indirect implication of subcellular accumulations of Na/Cl, with respect to the observed structural changes. The increase in the vacuole size was linearly related to Na/Cl concentration at leaf level (Fig. 7B, E; Table S1C) and the vacuolar allocation of contaminants indicated a direct salt accumulation effect (Fig. 4I, J). Given missing salt contaminants in cytoplasm (Fìg. 4I, J), the observed cell size increase with higher foliar salt concentrations (Fig. 7A, D; Table S1C) should be driven by that of vacuome, which was also confirmed by 1) the correlation between vacuome and cell size (P =0.03, r=0.57) and 2) concomitant thinning and degeneration of cytoplasmic strands (Fig. 6F, L, M), as a consequence of autophagic activity (Fig. 6I, J, L, M). As indicated by non-significant models for the cell circularity (Table S1C), the cell size increased both periclinally and anticlinally. Missing cytoplasmic accumulation of salt contaminants, the observed increase in the size of chloroplasts (Cl only) and plastoglobules (Na only) implied an indirect driving effect of salt contaminants (Fig. 7C, F; Table S1C). Only the models calculated using leaf rim data were significant; regarding those calculated on the basis of leaf center measurements, the cross-sectional area of palisade cells formed the best responsive estimate, increasing marginally (P =0.054; Cl) or as a tendency (P =0.088; Na) with rising salt accumulation in foliage.
Considering the cell responses in the palisade parenchyma from leaf rim samples globally, the salt contamination matrix (Na and Cl foliar concentrations) in the RDA model explained 78.12 % of observed variation in the structural markers from descriptor matrix (P=0.001; Fig. 8). However, only the first RDA axis (77.49 %) was significant (P =0.001). The two explanatory variables showed balanced contributions. The vacuole size (VS in Fig. 8) was the structural parameter most strongly contributing to RDA axis 1. The site centroids showed a distribution along first axis reflecting the gradient of average salt contamination in foliage. The passively projected macroscopic leaf injury parameter (Injury), which was significantly correlated with both Na and Cl (P < 0.001 and 0.002), showed a positive correlation tendency with respect to RDA first axis (P =0.073).