Aims In contrast to temperate regions, the reasons for growth reductions of individual tree species along elevational gradients in tropical mountain ranges are poorly known, especially for tropical conifers. We aimed at testing whether climatic or edaphic conditions are responsible for the reduced growth of Pinus kesiya, a widely distributed pine species of southern and south-eastern Asia, at higher elevations. Methods We analysed the stem diameter increment and the isotope discrimination against 13C (△ 13C) in tree rings of P. kesiya along an elevational gradient of ~900 to ~2000 m a.s.l. in the mountain ranges of South-Central Vietnam, and related growth to △  13C and to climatic and edaphic variables. Important findings We found no consistent correlation patterns between the basal area increment (BAI) of the trees and temperature or precipitation. In contrast, across the elevational gradient, we obtained significantly negative correlations of BAI with △  13C and with the C/N ratios and the δ 15N signature of the upper mineral soil. BAI was positively correlated with the concentrations of plant-available phosphorus (Pa) and of “base” cations (calcium, magnesium, potassium) in the soil. We conclude that lower temperatures at higher elevations exert an indirect effect on tree growth by inducing higher C/N ratios and by reducing the rate of nitrogen (N) and P mineralization, which may be further hampered by lower concentrations of “base” cations (upon enhanced leaching by precipitation) and a negative feedback from low availability of mineralized N and P at higher elevations. Our results may be transferable to the uppermost growth limit of P. kesiya and to other montane regions of the species’ occurrence.

Wood density constitutes an integrative trait of water relations and growth. We compared the recently developed blue intensity (BI) method, which has only rarely been applied to tropical conifers, for determining wood density with anatomical analyses in studying the three rarely investigated palaeotropical pine species Pinus kesiya, P. dalatensis and P. krempfii, which co-occur in South-Central Vietnam, but differ in their distribution areas. For species comparisons, we also calculated the hydraulic conductivity of the xylem with the Hagen-Poiseuille equation and the water potential causing 50% loss of hydraulic conductivity () based on the anatomical analyses. We hypothesized (i) that the BI values are correlated with the cell wall fractions, the calculated hydraulic conductivity and the values; and (ii) that the wider occurrence of P. kesiya, which also can grow at drier sites, is reflected by higher wood density, lower hydraulic conductivity, lower (more negative) values and a smaller variation in the wood anatomical features across the years compared to the other two species. In agreement to our hypotheses, the results of the BI and the anatomical method were closely correlated, especially for sapwood, and P. kesiya exhibited features that are related to the growth at drier sites and to a higher tolerance towards drought: higher wood density and cell wall:lumen area ratios of its smaller xylem conduits, lower calculated hydraulic conductivity and more negative values. The BI method is well suitable for determining the wood density in tropical conifers. As a fast and inexpensive method, it may be used for initial screening woody species for their water transport capacity and drought resistance.

Using a dendrochronological approach, we determined the resistance, recovery and resilience of the radial stem increment towards episodes of growth decline, and the accompanying variation of 13C discrimination against atmospheric CO2 (Δ13C) in tree rings of two palaeotropical pine species. These species co-occur in the mountain ranges of south–central Vietnam (1500–1600 m a.s.l.), but differ largely in their areas of distribution (Pinus kesiya from northeast India to the Philippines; P. dalatensis only in south and central Vietnam and in some isolated populations in Laos). For P. dalatensis, a robust growth chronology covering the past 290 years could be set up for the first time in the study region. For P. kesiya, the 140-year chronology constructed was the longest that could be established to date in that region for this species. In the first 40 years of the trees’ lives, the stem diameter increment was significantly larger in P. kesiya, but levelled off and even decreased after 100 years, whereas P. dalatensis exhibited a continuous growth up to an age of almost 300 years. Tree-ring growth of P. kesiya was negatively related to temperature in the wet months and season of the current year and in October (humid transition period) of the preceding year and to precipitation in August (monsoon season), but positively to precipitation in December (dry season) of the current year. The P. dalatensis chronologies exhibited no significant correlation with temperature or precipitation. Negative correlations between BAI and Δ13C indicate a lack of growth impairment by drought in both species. Regression analyses revealed a lower resilience of P. dalatensis upon episodes of growth decline compared to P. kesiya, but, contrary to our hypothesis, mean values of the three sensitivity parameters did not differ significantly between these species. Nevertheless, the vigorous growth of P. kesiya, which does not fall behind that of P. dalatensis even at the margin of its distribution area under below-optimum edaphic conditions, is indicative of a relatively high plasticity of this species towards environmental factors compared to P. dalatensis, which, in tendency, is less resilient upon environmental stress even in the “core” region of its occurrence.