RESULTS

Leaf phosphorus and nitrogen concentrations

The effect of P availability on P and N relations in leaves of B. attenuata and B. sessilis was tested by growing plants in sand, sand + laterite (SLAT) and sand + limestone (SLIM). Foliar total P concentrations on a mass basis for both B. attenuata and B. sessilis were highest when grown in sand, and lowest in SLIM (Fig. 1A). However, there were no significant differences between B. attenuata and B. sessilis in foliar total P concentrations for any substrate type (Fig. 1A). Mass-based foliar total P concentration for B. attenuata differed on all three substrates (P < 0.05): sand > SLAT > SLIM. In contrast to foliar total P concentration, foliar total N concentration on a mass basis for B. sessilis was greater than that for B. attenuata in each substrate type (Fig. 1B). The leaf total N concentrations on a mass basis for both species were the same in sand and SLAT, and higher than when grown in SLIM (P < 0.05).
The differences in mass-based foliar total P across substrates for bothB. attenuata and B. sessilis were also apparent when P concentrations were expressed on an area basis (P < 0.05, Fig. 1C). However, on an area basis, the foliar P concentration for B. attenuata was higher than for B. sessilis in all substrates. There were also substrate-dependent differences in area-based foliar total N concentration for B. attenuata , but not for B. sessilis(Fig. 1D). In contrast to higher mass-based foliar N concentration inB. sessilis than in B. attenuata in all substrates, the area-based foliar N concentration was higher in B. attenuata in SLAT, or was not distinguishable (Fig. 1D).
The concentration ratio of foliar total N to foliar total P (NTotal : PTotal) had the same pattern for the two species across the three substrate types (Table 1). The NTotal : PTotal ratio for both species was lower when grown in sand than when grown in SLAT or SLIM (P < 0.05). However, there was no significant difference for either species when grown in SLAT or SLIM. The NTotal : PTotal ratio in B. sessilis was significantly higher than that in B. attenuata in every substrate.

Photosynthetic rates, PPUE and PNUE

Net photosynthetic rates (Pn) for B. attenuata grown in SLIM were lower than for plants grown in sand or SLAT, while there were no substrate-dependent differences for B. sessilis (P< 0.05, Table 2). There were no differences in photosynthesis rate between the two species in any substrate (P > 0.05). Moreover, there were no substrate-dependent differences in PPUE within either species (Table 2). However, the PPUE of B. sessiliswas significantly higher than that of B. attenuata in all three substrates (P < 0.05, Table 2). The PNUE for B. attenuata grown in sand was higher than that of plants grown in SLIM (P < 0.05, Table 2), while PNUE for B. sessiliswas the same in all substrate types (Table 2). Moreover, there were no significant differences in PNUE between B. attenuata and B. sessilis for any substrate type (Table 2).

Leaf phosphorus fractions

Banksia attenuata and B. sessilis had different patterns of allocating leaf P to lipid, metabolite, nucleic acid and residual fractions in all substrates (Fig. 2). The lipid P concentrations of B. attenuatagrown in SLIM were lower than in plants grown in sand and SLAT (Fig. 2A), while the differences in metabolite P concentration in B. attenuata grown in the three substrate types was sand > SLAT > SLIM (Fig. 2B). The nucleic acid P concentrations ofB. attenuata grown in SLIM were lower than in plants grown in sand and SLAT (Fig. 2C). The lipid P, metabolite P, nucleic acid P and Pi concentrations of B. sessilis grown in sand were greater than those in plants grown in SLAT and SLIM (Fig. 2A, B, C, E).
The lipid P concentration of B. attenuata was greater than that of B. sessilis in SLAT and SLIM (Fig. 2A). The metabolite P and nucleic acid P concentrations of B. attenuata were lower than those of B. sessilis in both sand and SLIM, but were the same in SLAT (Fig. 2B, C). The residual P concentration only differed between the species when grown in SLAT, with the concentration in B. attenuata being lower than that in B. sessilis (Fig. 2D). The Pi concentrations of B. attenuata were greater than those ofB. sessilis in sand and SLAT, but not SLIM (Fig. 2E).
The NTotal : PFraction ratios for metabolite P and Pi for B. attenuata were significantly lower in sand than in SLAT or SLIM, while the ratios for the other fractions were indistinguishable among the substrates (Table 1). This result shows that metabolite P and Pi fractions were drivers for the observed difference in leaf NTotal : PTotal ratio forB. attenuata in the three substrates. For B. sessilis , all P factions except residual P contributed to the lower leaf NTotal : PTotal ratio in sand. The ratios of leaf NTotal : PFraction in each P fraction were lower in B. attenuata than in B. sessilis for plants grown in any of the three substrates (Table 2).

The proportions of foliar P fractions to total foliar P

The proportions of lipid P and nucleic acid P in B. attenuatawere significantly lower for plants grown in sand than when grown in SLAT or SLIM (P < 0.05, Fig. 3A, C). Conversely, the proportion of total P in Pi for B. attenuata was greater for plants grown in sand than for plants grown in SLAT or SLIM (P< 0.05, Fig. 3E). There was no significant difference in the proportions of total P in lipids , metabolites, nucleic acid and Pi for B. sessilis grown in any of the three substrates (Fig. 3A).
The proportion of total P in lipid P in B. attenuata was greater than that of B. sessilisin SLAT and SLIM (P < 0.05, Fig. 3A). Conversely, the proportion of total P in nucleic acid P in B. sessilis was significantly greater than that in B. attenuata in all substrates (P < 0.01); likewise, the proportion of metabolite P inB. sessilis was greater than in B. attenuata in SLIM (Fig. 3B, C). The proportion of residual P was below 10% in all substrates, except for B. sessilis grown in SLAT (Fig. 3D). The proportion of total P in Pi in B. attenuata was greater than that of B. sessilis in all three substrates (P < 0.05, Fig. 3E).
The relationships of RGR and foliar nutrient concentrations, foliar P fractions, and leaf mass per area
The RGR of B. attenuata was lower than for B. sessilis in all substrates (Table 1). In B. attenuata, the RGR was the same in sand and SLAT, but lower in SLIM, while in B. sessilis, RGR was also highest in sand, but lower and equal in both SLAT and SLIM. The RGR of B. attenuata and B. sessilis was positively correlated with foliar nucleic acid P concentration (Fig. 4A) and foliar N concentration (Fig. 4B). There was also a positive correlation between foliar N concentration and nucleic acid P concentration for both species (Fig. 4C). No other correlations were detected in any other pairwise comparison of P concentration with N concentration. In both species, the concentration P in lipid, metabolite, nucleic acid, residual and Pi fractions were all correlated positively with foliar total P concentration (Fig. 5).
Leaf mass per area varied from 240 to 267 g m-2 in B. attenuata and 119 to 131 g m-2 in B. sessilis depending on the growth substrate (Fig. 5, Table 2). The concentrations of the P fractions and total P generally decreased with increasing LMA (Fig. 5). This relationship was not significant for lipid P and nucleic acid P inB. sessilis . The only exception to a negative correlation between leaf P fraction and LMA was the lack of a relationship between residual P and LMA in B. attenuata .