Study site
Our research was conducted in the Pasoh Forest Reserve (latitude 2°58′N, longitude 102°18′E, elevation 75–150 m a.s.l.), Negri Sembilan, Peninsular Malaysia. The forest reserve is covered with a primary lowland forest, with Dipterocarpaceae species predominating (Niiyama et al. 2010). The geography is dominated by flat alluvial plains interspersed with tiny swales, riverine areas, and gently undulating hills (Manokaran et al. 2004). Ultisol is the predominant soil type in the study area (Soil Survey Staff 2006), which forms primarily from parent rocks such as shale, granite, and fluviatile granite alluvium (Allbrook 1973). The A horizon is found at a depth of 0–5 cm, while lateritic gravels are prevalent below 30 cm (Yamashita et al. 2003). Between 1997 and 2011, the mean annual air temperature was 25.4ºC, while the mean annual precipitation in 1995–2015 was 1805 mm (Noguchi et al. 2016; Lion et al. 2017). Furthermore, the precipitation peaks at March–May and October–December, and the rainfall mostly occurs from late afternoon to night (Kosugi et al. 2008).
Experimental design
We targeted five Macaranga species (M. conifera, M. gigantea, M. hypoleuca, M. lowii, and M. recurvata) and eight Shorea species (S. acuminata, S. leprosula, S. macroptera, S. maxwelliana, S. multiflora, S. ovalis, S. parvifolia, and S. pauciflora), all of which grow in the lowland dipterocarp forest. In 2014 and 2015, we investigated the habitat environment and root tip morphology of these species. For the habitat environment survey, we randomly selected ≥ 7 individuals of < 4 m height (Table 1) for each target species in the vicinity of the International Biological Program's 6 ha plot (Ashton et al. 2003). For the root tip morphology survey, we randomly selected ≥ 6 individuals of < 4 m height from each species for the study of root tip morphology because some individuals died following the habitat environment survey.
Survey of habitat environment
We evaluated light conditions, soil physical properties, and soil nitrogen (N) dynamics surrounding target individuals to determine how habitat environment links resource acquisition and utilization by target species. To assess the light levels on a target individual, we measured Gap Light Index (GLI), which indicates the percentage of photosynthetically active radiation transmitted through a gap to a point in the understory. In September 2014, we took hemispherical photos at 3 points on the crown of a target individual using a digital camera (D5200, Nikon, Tokyo, Japan) equipped with a fisheye lens (Circular fisheye 4.5 mm, Sigma, Kawasaki, Japan). The hemispherical photo is then converted to a black-and-white image using a software called Side Look (version 1.1: Nobis 2005; Nobis and Hunziker 2005; Glatthorn and Beckschäfer 2014), and the GLI is calculated using a software called Gap Light Analyzer (version 2.0: Frazer et al. 1999).
We evaluated soil bulk density and sand content because they are connected to soil compaction and water permeability, respectively. In September 2014, we collected 200 cm3 soil samples from 0 to 10 cm depth in three points surrounding each target individual using a cylindrical soil sampler with a bottom area of 20 cm2 and a height of 5 cm (Stainless sampling cylinder 100 mL, Daiki Rika Kogyo, Kounosu, Japan). To remove plant tissues and gravels from the soil sample, it was put through a 2-mm mesh screen. The sieved soil was then oven dried for 24 hours at 105°C and weighed. The bulk density of soil was calculated by dividing the dry weight of sieved soil by the sample volume (i.e., 200 cm3). Additionally, we mixed the three sieved soil samples collected around a target individual and measured the sand content of the mixed sample; we eliminated organic matter from the mixed sample using H2O2, and used sodium hexametaphosphate to disperse the soil particles. The size distributions of soil particles were determined by the ISSS fractionation system (Gee and Or 2002). We determined the sand content by calculating the percentage of fine and coarse sands in the dry weight of sieved soil.
For soil N dynamics, we examined the net ammonification rate (hereafter referred to as ammonification rate), the net nitrification rate (hereafter referred to as nitrification rate), and the net N mineralization rate (hereafter referred to as N mineralization rate), all of which are related to the supply of available N. In December 2014, we collected 200 cm3 of soil from 0 to 10 cm depth in three locations around a target individual using a cylindrical soil sampler with a bottom area of 20 cm2 and a height of 5 cm (Stainless sampling cylinder 100 mL, Daiki Rika Kogyo, Kounosu, Japan). We combined the three soil samples, transported them to the laboratory in a cooler box, and refrigerated them. Within 48 hours of sampling, the mixed sample was sieved with a 2-mm mesh screen to remove plant tissues and gravels. We employed sieved soil to measure the ammonification and nitrification rates; we determined the amount of ammonium N and nitrate N in the sieved soil before and after incubation. To determine pre-incubated levels of ammonium N and nitrate N, we added 50 mL of 2 M KCl to 5 g of sieved soil (in wet weight) immediately after sifting to extract inorganic N. The suspension was agitated vigorously for one hour and then filtered using filter paper (Quantitative filter paper No 6, Tokyo Roshi Kaisha, Ltd., Tokyo, Japan). The ammonium N and nitrate N contents in the filtrate were measured using an autoanalyzer (FIAstar5000、FOSS、Hillerød、Denmark). Simultaneously, we oven dried another portion of the sieved soil for 24 hours at 105°C and weighed it to obtain the gravimetric soil water content, which was used to calculate the ammonium N and nitrate N contents per soil dry weight. For the post-incubated ammonium N and nitrate N, 5 g of sieved soil (in wet weight) was incubated in a polyethylene container covered with parafilm with tiny pores and kept at 28°C for 28 days in the dark. The moisture content of the incubated soil was maintained at 60% of the maximal water retention capacity of typical soil samples by weekly addition of deionized water. Following incubation, we determined the amount of ammonium N and nitrate N in the soil dry weight using the same methods as for pre-incubated values. The ammonification and nitrification rates were determined by dividing the difference between the pre- and post-incubation values by the number of days of incubation (i.e., 28 days). Additionally, we computed the rate of N mineralization as the sum of the ammonification and nitrification rates.
For statistical analysis, the mean values of GLI at three points on the crown of a target individual and the mean values of soil bulk density at three points around the target individual were used. For the sand content, ammonification rate, nitrification rate, and N mineralization rate, statistical analysis was performed using a mixed sample measurement value.
Survey of root tip morphology
In August 2015, the roots of a target individual were traced from the stem and sampled to examine the root tip morphology. We took ≥ 2 samples from each target individual; however, for very small individuals, we took one sample of as much fine root as feasible, i.e., whole root system. Within 24 hours of collection, these samples were transported to the laboratory, and soil adhering to fine roots was carefully cleaned under running water. These root samples were stored in the freezer until the root tip morphology was determined.
In September 2015, we determined the specific root tip length (SRTL), root tip diameter (RTD), and root tip tissue density (RTTD) of these root samples. After defrosting the preserved fine roots in water, the first-order roots, i.e., root tips, were severed from the fine root system in accordance with Pregitzer et al. (2002). The root tips were scanned with a scanner (GT-X070, Epson, Nagoya, Japan), and the root tips' RTD, length, and volume were determined using the image analysis software WinRhizo (Win RHIZO regular, Regent Instruments, Quebec, Canada). The root tips were dried at 70°C for 48 hours, and the dry weight was measured. The SRTL was computed from this data by dividing the root tip length by their dry weight, and the RTTD was calculated by dividing the root tip dry weight by the volume of the root tip.
Statistical analysis
To confirm the phylogenetic effect on habitat environment, we used a generalized linear mixed model (GLMM) of Formula 1 to compare the GLI, soil bulk density, sand content, ammonification rate, nitrification rate, and N mineralization rate of the two genera, Macaranga and Shorea. In Formula 1, we considered tree species as a random effect because species are assumed to be nested within each genus. Moreover, to check the difference in habitat environment among the target species, we analyzed the effect of species on the variation of GLI, soil bulk density, sand content, ammonification rate, nitrification rate, and N mineralization rate using a GLMM of Formula 2. In Formula 2, we considered genus as a random effect to cancel out phylogenetic effect. In both Formula 1 and 2, we assumed a gamma distribution with a log link function for the GLI, soil bulk density, sand content, nitrification rate, and N mineralization rate, and a Gaussian distribution for the ammonification rate. We used the Type II test of analysis of deviance to determine the significance of the fixed effect, i.e., genus in Formula 1 and species in Formula 2. Additionally, a multiple comparison was performed to determine the difference between each pair of species using the GLHT function in conjunction with the Tukey method based on a GLMM of Formula 2.
[Formula 1] Environmental factor = Genus + (Species)
[Formula 2] Environmental factor = Species + (Genus)
To quantify the variation in combined environmental factors, we conducted Principle Component Analysis (PCA) using the six factors: GLI, soil bulk density, sand content, ammonification rate, nitrification rate, and N mineralization rate.
To confirm the phylogenetic effect on root tip morphology, we used a GLMM of Formula 3 to assess the difference in SRTL, RTD, and RTTD between the two genera. In Formula 3, we considered tree species as a random effect because species are assumed to be nested within each genus. Additionally, to corroborate the difference in root tip morphology among target species, we examined the effect of species on SRTL, RTD, and RTTD variation using a GLMM of Formula 4. In Formula 4, we considered genus as a random effect in order to balance out the phylogenetic effect. In both Formula 3 and 4, a gamma distribution with a log link function was assumed. We used the Type II test of analysis of deviance to determine the significance of the fixed effect, i.e., genus in Formula 3 and species in Formula 4. Additionally, a multiple comparison was performed to ascertain the difference between each pair of species using the GLHT function in conjunction with the Tukey method based on a GLMM of Formula 4.
[Formula 3] Root tip morphology = Genus + (Species)
[Formula 4] Root tip morphology = Species + (Genus)
To better understand the association between root tip morphology and habitat environment, we used a generalized linear model (GLM) of Formula 5 to compare the change in SRTL, RTD, or RTTD to the scores on each axis of PCA (PC1–3). Additionally, to validate the association between root tip morphology and a specific environmental factor, we used a GLM of Formula 6 to assess the change in SRTL, RTD, or RTTD with GLI, soil bulk density, sand content, ammonification rate, nitrification rate, and N mineralization rate. In Formula 5 and 6, we considered the phylogenetic effect (i.e., genus) as a fixed effect in order to find changes in root tip morphology variation between genera. Formulas 5 and 6 assumed a gamma distribution with a log link function. We determined the significance of the fixed effect in Formula 5 and 6 using the Type II test of analysis of deviance.
[Formula 5] Root tip morphology = Principal component scores + Genus + Principal component scores × Genus
[Formula 6] Root tip morphology = Environmental Factor + Genus + Environmental factor × Genus
All statistical analyses were conducted in R4.1.3 (R Core Team 2021).