Variation in soil phosphorus (P) availability promotes niche differentiation among tropical tree species, yet the traits that underpin specialization on low-P and high-P soils (hereafter low-P and high-P species) remain poorly understood. Here, we examined interspecific and intraspecific variation in three types of root phosphatase enzymes and morphological traits among neotropical tree species in Panama with different habitat associations. We collected fine roots from 51 individual trees of four congeneric pairs of low-P and high-P neotropical tree species in the genera Cordia, Hirtella, Inga, and Protium in forests on moderate to low-P soils. We determined root morphological traits (specific root length, diameter, and root tissue density), and the root-surface activities of phosphomonoesterase (PME), phosphodiesterase (PDE), and phytase (PHY) enzymes, which are synthesized to release inorganic orthophosphate from soil organic P. Soil P availability was determined by measuring resin-extractable P concentration for soils collected from the base of each tree. Low-P species allocated more resources to produce enzymes that decompose more complex forms of P, as indicated by greater PHY activity, and greater PHY:PME and PDE:PME ratios at a given soil P availability. A principal component analysis of fine-root traits showed a greater Euclidian distance among individuals of low-P species than among those of high-P species, supporting the hypothesis that fine-roots traits vary more among low-P species than among high-P species.

Synthesis. These results suggest that the specialization of tropical tree species to low-P soil involves investment in the acquisition of complex soil organic phosphates such as phosphodiesters and phytic acid. This is possibly related to root trait divergence and indicates that variation in P acquisition strategies among tropical tree species could contribute to resource partitioning on low-P soils.

Study site and species

We sampled leaves and fine roots from trees in three lowland forest census plots in the Panama Canal watershed (Pyke et al. 2001, Turner and Engelbrecht 2011, Condit et al. 2013). Vegetation is semi-deciduous, seasonally moist forest, with 2311-2848 mm of annual rainfall across the four plots and a mean annual temperature of 27°C. The four sites represent moderate (Barro Colorado Island) to low (Pipeline Road, Fort Sherman) soil P availability in this region (Table 1), based on a 2 mg P kg^–1 threshold at which there are marked differences in the response to P, including growth, distribution of low-P and high-P species, and soil phosphatase (Condit et al. 2013; Turner et al. 2018). Condit et al. (2013) showed that regional tree species distributions are structured by gradients of moisture and P. We selected moist forests to minimize effects of moisture availability on species traits. High-P and low-P species were determined by analyzing the responses of individual tree species to multiple environmental factors including soil P and moisture using the data of 72 sites and 550 species across Panama (Condit et al. 2013). This yielded  effect sizes for each species: species with P effect sizes < 0 had a negative relationship between occurrence frequency and soil P availability (i.e., low-P species) (Table 2). A pair of low-P and high-P species were selected for each of Cordia (C. bicolor and C. alliodora; Verbenaceae), Hirtella (H. americana and H. triandra; Chrysobalanaceae), Inga (I. pezizifera and I. nobilis; Fabaceae), and Protium (P. panamense and P. tenuifolium; Burseraceae). The selected species are widespread subcanopy tree species in Panamanian neotropical forests.

Field sampling

We measured the activity of three phosphatases (PME, PDE; and PHY), mycorrhizal colonization, and root morphological traits. We included mycorrhizal colonization rate because it might affect plant P acquisition and phosphatase activities. At least six understory individuals were identified for each species, with diameter at breast height (DBH) between 50 and 300 mm. Fine roots (i.e., roots < 2-mm diameter) of each individual tree were sampled by tracing coarse roots from the trunk. Roots were immediately chilled in the field and transferred to the laboratory. We collected several root branches to determine physiological and morphological traits. Some individuals had few fine roots that could be tracked, so data for some parameters were incomplete, because we prioritized phosphatase activity data over mycorrhizal colonization data (e.g., for Cordia bicolor) and therefore omitted the data of mycorrhizal colonization from some analyses (see below). We collected bulk soil at 0–10 cm depth beneath the litter layer around each individual tree to determine local P availability. Finally, we collected shade leaves from the canopy of each tree to calculate leaf mass per area (g m^–2) as an indicator of above-ground nutrient-use strategy.

Laboratory analyses

We measured root phosphatase activity (PME, PDE, and PHY), mycorrhizal colonization rate, and morphological traits (root diameter, specific root length, and root tissue density) for each fine-root sample and P availability for each soil sample. We determined fine-root PME activity using a modified method of Yokoyama et al. (2017). Each of two 10-mg subsamples of fine-root tissue was incubated with 2.25 ml of sodium acetate-acetic acid buffer adjusted to pH 5.0 and 0.25 ml of 25 mM para-nitrophenyl phosphate (pNPP) for 30 minutes in a water bath at 26 °C (the mean soil temperature in lowland tropical forests of Panama). The buffer included 3 mM NaN₃ to inhibit microbial activity. We also prepared blanks that received buffer and substrate without fine roots. The enzymatic reaction was terminated by adding 0.5 mL of the assay solution to 4.5 mL of 0.11 M NaOH. The concentration of para-nitrophenol (pNP) was determined by measuring the absorbance at 405 nm on a spectrophotometer, after subtracting absorbance in blanks (roots but no substrate) and controls (substrate but no roots). Phosphodiesterase activity was determined by the same procedure, except using 25 mM bis-para-nitrophenyl phosphate (bis-pNPP) as substrate.

The measurement of PHY activities was based on a modified method of Yokoyama et al. (2017). Five 10-mg subsamples of fine-root material were weighed into glass vials (indicated as S1, S2, C, P, N for each subsample). We also prepared glass vials without roots (indicated as B sample: see the later explanation for what these subsamples represent). Subsamples S1, S2, and B received 0.25 ml of 8 mM phytic acid (sodium salt) in acetate buffer (pH 5.0). Subsample C and N received 0.25 ml of acetate buffer. Subsample P received 0.25 ml of 2 mM KH₂PO₄. The subsamples were incubated for 14 h in a water bath at 26 °C and enzymatic reaction was terminated by adding 1 ml of 10% trichloroacetic acid (TCA). Subsequently, subsamples S1, S2, P, N received 1 ml of acetate buffer whereas subsample C received 0.25 ml of 2 mM phytic acid sodium salt in acetate buffer. Incubated subsamples were centrifuged, and orthophosphate was determined in the supernatants by molybdate colorimetry with the absorbance (712 nm) on a spectrophotometer using a 1-cm cell. PHY activity was calculated by the following equation:

PHY activity = (S–C–B)/[T*(P–N)/P_b] (µmol PO₄ g^–1 hr^–1), (Equation 1)

where S is the mean PO₄ of subsamples S1 and S2 after incubation per subsample root weight (μmol PO₄ g^-1); C indicates PO₄ in subsample C after incubation per subsample root weight (μmol PO₄ g^-1); P indicates PO₄ concentration in the subsample P after incubation (μg PO₄ mL^-1); N indicates PO₄ concentration in the subsample N after incubation (μg PO₄ mL^-1); T indicates Incubation time (hr); Pb indicates the added PO₄ concentration in the subsample P before incubation (μg PO₄ mL^-1). Subsample C, B, and (P –N)/P_b represent PO₄ release from fine roots during incubation, PO₄ release from phytic acid sodium salt, and collection rate of PO₄, respectively.

The intensity of mycorrhizal colonization was determined by a modified method of Giovannetti and Mosse (1980). Root branches were cleaned with deionized water, soaked in 10% KOH in a 95°C water bath for 3 min, acidified with 2% HCl solution for 1 h, and stained with 0.02% Trypan blue in a 90°C water bath for 30 min, after which the stained root branches were placed in a solution of lactic acid, glycerol and water to remove excess stain (decolorization). If roots still had a dark color after boiling with KOH solution, they were soaked into 10% H₂O₂ and 20% NH₄OH for 10 min. The degree of root mycorrhizal colonization was quantified based on the method described by McGonigle et al. (1990).

In brief, the cleaned and stained root branches were divided into fragments (ca. 1 cm) and mounted on slides. Root fragments were observed at 1-mm intervals at ×200 magnification using a microscope (100 to 200 points), and the proportion of root length containing fungal structure (arbuscules, vesicles and hyphae) was calculated.

For morphological traits, roots were scanned at 300 dpi and average diameter and length determined using a semiautomated image-analysis software, Smart Root (Guillaume et al. 2011) based on Image J (Schneider et al. 2012). We manually tracked fine-roots used for PME analysis, calculating root length and average root diameter. All root branches were oven-dried at 65°C for two days and weighed. Specific root length and root tissue density were calculated as total length divided by dry mass, and dry mass divided by total volume, respectively. Leaves were scanned at 300 dpi and areas determined with Image J. Leaves were oven-dried at 65°C and weighed. Leaf mass per area was calculated as dry weight divided by leaf area.

Readily available P was determined using anion-exchange membranes (hereafter “resin P”). Samples were analyzed immediately after soil collection, to eliminate artifacts caused by soil storage (Turner and Romero 2009). Briefly, 5-g soil was shaken for 24 h with 80 mL of deionized water and five anion-exchange membrane stripes (1 cm × 4 cm). The strips were rinsed in deionized water and then shaken for 1 h in 50 mL of 0.25 M H₂SO₄. Phosphate was determined in the extracts by automated molybdate colorimetry on a Lachat Quikchem 8500 (Hach Ltd, Loveland, CO).