Aim Warm deserts are characterized by water shortages and high-temperature extremes. A common adaptive strategy in such environments is the maximization of photosynthetic capacity, which allows plants to achieve positive carbon budgets by taking advantage of short periods of water availability and non-inhibitory temperatures. Assuming that photosynthetic capacity is approximated by leaf N concentration, we tested the hypothesis that environmental aridity is related to an elevated leaf nitrogen content.

Location 53 locations in the transitional zone spanning the Mediterranean and the Sahara Desert in Morocco. The mean maximal temperature (T_max) within the area varied between 35.7 and 43.5 °C, and the mean annual precipitation (MAP) was between 12 and 246 mm.

Taxon 225 vascular species representative of local vegetation

Methods Leaf samples were collected along a regional aridity gradient and preserved in herbarium presses. The leaf mass per area (LMA) and N concentrations expressed on leaf mass (N_mass) and area (N_area) basis were determined. We also obtained LMA and N_mass values for 6711 species from a worldwide database for comparative analysis.

Results Significant increases in mean LMA, N_mass, and N_area accompanied the increase in T_max and the decrease in MAP in woody species and in non-graminoid herbs, but not in graminoids. Considering that the climate in our sampling area as a whole was arid, we compared the N_mass values of Moroccan plants with those from a worldwide database and found that at a common LMA, the Moroccan plants showed on average elevated N_mass relative to the global values.

Main conclusions These two lines of evidence: regional gradient and global comparison confirm that hot deserts select for high leaf N content. This result predicts the direction of natural selection that will accompany future climate warming.

Field sampling - Plant material was collected during two field trips conducted between 9 and 18 June 2011 and between 12 and 22 March 2012 at 53 sites located within the state limits of Morocco. Site selection was haphazard and was based on the presence of relatively undisturbed spontaneous vegetation. One to several species were sampled per site. Two aggregate samples consisting of several healthy, mature, well-illuminated leaves per species were collected, each from several separate individuals. The total number of taxa sampled (sometimes identified down to subspecific level but hereafter referred to as species) was 225; however, 22 species were sampled from two sites and one from three sites. The leaves were carefully placed between sheets of newsprint and blotting paper and pressed tightly in a herbarium press. The blotting paper was exchanged daily for several days until the material was dry.

Sample processing - Samples were transported to the laboratory, where areas of individual leaf laminas were measured using a desktop scanner controlled by WinFOLIA software (Regent Instruments, Sainte Foy, Quebec, Canada). Typically, 5 or 10 leaves per sample were used, but in several species, the number was lower (2-4). The leaves were then placed in a forced circulation oven at 65 °C for 72 hours and weighed, and leaf mass-per-area (LMA, g m^-2) was calculated. Leaf material was ground in a mill (IKA Labortechnik, Staufen, Germany) and leaf N content per mass (N_mass) was analyzed in an Elemental Combustion System CHNS-O 4010 (Costech Instruments, Italy/USA). Leaf N content per area (N_area) was calculated by multiplying N_mass by LMA.

Evaluation of the press drying technique - To assess the effect of using desiccated material on the accuracy of trait determination, a set of mature leaves from diverse species widely varying in shape and structure was collected in August 2018 from the grounds of Adam Mickiewicz University campus in Poznań, Poland. Thirty-nine species were sampled and the number of leaves per species was 14-20. In Picea omorica several needles constituted a single sample. The samples were placed in plastic bags containing moist paper towels and brought to the laboratory, and the areas of fresh leaf laminas were measured with a desktop scanner and Winfolia software. Within each species, samples were randomly assigned to one of two drying treatments. The first treatment involved immediate drying of the leaves in a forced circulation oven at 65 °C for 72 hours. For the second treatment, the leaves were placed in herbarium presses, as described above. The drying in the presses lasted four weeks. The blotting paper was replaced daily during the first week and less frequently thereafter. The presses were kept indoors in a well-ventilated spot with variable temperatures that on most days exceeded 30 °C. When drying in the presses was complete, leaves were again scanned to determine leaf areas followed by drying at 65 °C for 72 hours. All dry leaves were weighed and LMA was calculated. For N determination, four press-dried and four oven-dried leaf samples from each species were ground and analyzed as described above.