Metrosideros polymorpha, a dominant tree species in the Hawaiian Islands, shows an extreme phenotypic polymorphism both across gradients of climatic/edaphic conditions and within populations, making it a potentially useful model species for evolutionary study. In order to understand how the phenotypic diversity is maintained within populations as well as across populations, we examined the diversities of several leaf and stem functional traits across five elevations and two soil substrates on the volcanic mountain of Mauna Loa, on the island of Hawaii. Leaf dry mass per area (LMA), a key leaf functional trait, was particularly focused on and analyzed in relation to its underlying components-namely, tissue LMA and trichome LMA (LMA = tissue LMA + trichome LMA). Across populations, tissue LMA increased linearly with elevation while trichome LMA showed unimodal patterns with elevation, which were better correlated with temperature and rainfall, respectively. Substantial phenotypic variations were also found within populations. Interestingly, the variations of tissue LMA were often negatively correlated to trichome LMA within populations, which contrasts with the cross-populations pattern, where a strong positive correlation between tissue LMA and trichome LMA was found. This suggests that phenotypic variations within populations were substantially influenced by local ecological processes. Soil depth (an indicator of local water availability) and tree size (an indicator of colonized timing) modestly explained the within-population variations, implying other local environmental factors and/or random processes are also important in local phenotypic diversity. This study provides an insight about how phenotypic diversity of plant species is maintained from local to landscape levels.

Nutrient resorption, a process by which plants degrade organic compounds and resorb their nutrients from senescing tissues, is a crucial plant function to increase growth and fitness in nutrient-poor environments. Tropical trees on phosphorus (P)-poor soils are particularly known to have high P-resorption efficiency (PRE, the percentage of P resorbed from senescing leaves before abscission per total P in green leaves). However, the biochemical mechanisms underlying this greater PRE remain unclear. In this study, we determined the P concentration in easily soluble, nucleic acid, lipid and residual fractions for green and senescent leaves of 22 tree species from three sites, which differed in P availability, on the lower flanks of Mt. Kinabalu, Borneo. PRE varied from 24 to 93% and was higher in species from the P-poor site. P-resorption rate was greatest from the lipid fraction, the nucleic acid fraction, and lowest in the easily soluble fraction and the residual fraction when all the species were pooled. For species with higher PRE, P-resorption rate of the residual fraction was relatively high and was comparable in magnitude to that of the other labile fractions. This suggests that tree species inhabiting P-poor environments increased PRE by improving the degradation of recalcitrant compounds. This study suggests that plants selectively degrade organic compounds depending on environmental conditions, which is a key mechanism underlying the variation of PRE.

Summary1. Tropical rain forests in SE Asia are well known for the occurrence of supra-annual synchronous reproductive events, masting. Answering the question how trees allocate carbon (C), nitrogen (N) and phosphorus (P) to such irregular but gregarious reproduction requires a long-term observation. We conducted a 10-year continuous monitoring of litterfall in eight tropical rain forests, which differ significantly in P (and N) availability on Mount Kinabalu, Borneo. 2. Mean P concentration in reproductive organ litter decreased significantly with increasing P-use efficiency of net primary production (PUE), an index of P deficiency. Therefore, P in reproductive organ litter became diluted as the magnitude of P deficiency increased. Mean annual litterfall (kg haÀ1 yr À1 ) of reproductive organs over the 10 years ranged from 128.5 to 730.9 across the eight forests. Long-term C allocation ratio to reproductive organs (i.e. C in reproductive organs per C in total litterfall) varied from 2.0% to 7.8% across the eight forests and did not relate with PUE, indicating that long-term C allocation ratio to reproduction was not controlled by the availability of the most critical soil nutrient. 4. Long-term N allocation ratio to reproduction varied from 2.7% to 9.9% and significantly positively related with C allocation ratio. The quotient of N allocation ratio to C allocation ratio ranged from 1.1 to 1.4. Long-term P allocation ratio to reproduction varied from 9.8% to 16.4%. The quotient of P allocation ratio to C allocation ratio ranged from 1.6 to 5.0. Therefore, tropical trees allocated much greater proportion of P to reproduction than C and N over the 10 years. Moreover, trees disproportionately increased P allocation to reproduction with decreasing C allocation to reproduction. Trees adjusted P allocation relative to C allocation and maintained a narrow range of P allocation ratio to reproduction in the long run in each site. 5. Synthesis. Reproduction in Bornean tropical rain forests costs more P than C and N. Our results suggest that reproductive events in these forests are regulated by P at the level of overall long-term mean. Understanding patterns and processes of reproductive events requires a long-term monitoring of nutrient dynamics.

Genomewide markers enable us to study genetic differentiation within a species and the factors underlying it at a much higher resolution than before, which advances our understanding of adaptation in organisms. We investigated genomic divergence in Metrosideros polymorpha, a woody species that occupies a wide range of ecological habitats across the Hawaiian Islands and shows remarkable phenotypic variation. Using 1659 single nucleotide polymorphism (SNP) markers annotated with the genome assembly, we examined the population genetic structure and demographic history of nine populations across five elevations and two ages of substrates on Mauna Loa, the island of Hawaii. The nine populations were differentiated into two genetic clusters distributed on the lower and higher elevations and were largely admixed on the middle elevation. Demographic modelling revealed that the two genetic clusters have been maintained in the face of gene flow, and the effective population size of the high-altitude cluster was much smaller. A F -based outlier search among the 1659 SNPs revealed that 34 SNPs (2.05%) were likely to be under divergent selection and the allele frequencies of 21 of them were associated with environmental changes along elevations, such as temperature and precipitation. This study shows a genomic mosaic of M. polymorpha, in which contrasting divergence patterns were found. While most genomic polymorphisms were shared among populations, a small fraction of the genome was significantly differentiated between populations in diverse environments and could be responsible for the dramatic adaptation to a wide range of environments.

The residence time of phosphorus (P) in trees is a consequence of plant adaptation to P deficiency, with longer P residence time on soils with low P availability. P residence time has been studied at the leaf or canopy level but seldom at the whole‐tree level. Whereas P residence time at the leaf or canopy level is largely determined by leaf longevity and the resorption of P before leaf abscission, P residence time at the whole‐tree level will also be influenced by differences in P allocation to different plant parts because leaves and woody organs have distinct longevities. We estimated the residence time of P in above‐ground tree biomass (AGB) as the ratio of P mass (i.e. leaves plus wood) to the annual flux of P via litterfall (i.e. fine litter plus coarse woody debris) for seven tropical rain forests with different soil P availabilities on Mount Kinabalu, Borneo. We analysed the effects of P allocation to and resorption from leaves on P residence time along a soil P gradient. P residence time (2.7–9.8 years) was approximately one‐fifth of biomass residence time (AGB/annual litterfall mass; 19.8–48.8 years). This was due to a disproportionately greater relative allocation of P to leaves (P mass in leaves/P mass in AGB; 0.11–0.46), which had a smaller fraction of biomass (leaf biomass/AGB; 0.02–0.05) but a shorter longevity (1.0–1.8 years). The relative allocation of P to leaves was often high on low‐P soils, and P residence time was expected to be short. By contrast, the resorption rate of P from leaves was also high on low‐P soils, which extended P residence time with P deficiency. Consequently, P residence time was nearly constant across the forests. The short residence time of P relative to biomass indicates that P residence time depends largely on relative P allocation among plant organs. Similar P residence times among sites were maintained because greater P allocation to leaves on low‐P soils was effectively offset by higher P‐resorption efficiency. A free Plain Language Summary can be found within the Supporting Information of this article.