High nutrient-use efficiency during early seedling growth in diverse Grevillea species (Proteaceae)

Several hypotheses have been proposed to explain the rich floristic diversity in regions characterised by nutrient-impoverished soils; however, none of these hypotheses have been able to explain the rapid diversification over a relatively short evolutionary time period of Grevillea, an Australian plant genus with 452 recognised species/subspecies and only 11 million years of evolutionary history. Here, we hypothesise that the apparent evolutionary success of Grevillea might have been triggered by the highly efficient use of key nutrients. The nutrient content in the seeds and nutrient-use efficiency during early seedling growth of 12 species of Grevillea were compared with those of 24 species of Hakea, a closely related genus. Compared with Hakea, the Grevillea species achieved similar growth rates (root and shoot length) during the early stages of seedling growth but contained only approximately half of the seed nutrient content. We conclude that the high nutrient-use efficiency observed in Grevillea might have provided a selective advantage in nutrient-poor ecosystems during evolution and that this property likely contributed to the evolutionary success in Grevillea.

Species in the family Proteaceae are highly adapted to and became diversified on the most nutrient-impoverished soils observed worldwide. Highly infertile soils have long been demonstrated to influence biological and physiological responses and speciation patterns 4,9 . Species have developed diverse morphological and physiological adaptations to limited nutrient availability in order to successfully establish in nutrient-limited soils [10][11][12][13] . These species typically exhibit highly efficient photosynthetic phosphorus use 14 .
In southwestern Australia, seed germination and successful seedling establishment predominantly occurs following winter rains after summer fires. Seed-stored nutrients, particularly nitrogen (N) and phosphorus (P), are crucial for the emergence and early growth of seedlings, and the efficient use of these nutrients is critical for the successful establishment of seedlings in severely nutrient-poor and water-limited environments. Nutrient limitation has recently been proposed as a potential driver for variation in species diversity 15 . Lambers et al. suggested that high plant species diversity on infertile soils, particularly in species-rich biomes, is associated with functional diversity for nutritional strategies, including nutrient-use efficiency 16 . Despite extensive research on plant species richness in relation to nutrient availability, few studies have attempted to explore the relationship between nutrient-use efficiency and speciation patterns. Here, we compare the nutrient concentration and content in seeds and nutrient-use efficiency during early seedling growth of Grevillea with those of the closely related genus Hakea, another Australian genus with 149 species that efficiently use seed nutrients during early seedling growth 16 . Our study aimed to provide unique insights into the potential mechanisms underlying the evolutionary success (i.e., high diversification rate) of Grevillea.

Results
Seed mass varies markedly in both Grevillea (2.8-244 mg, 116 species/subspecies surveyed) and Hakea (2.6-501 mg, 153 species/subspecies surveyed), and Grevillea has a slightly higher median seed mass than Hakea (24.2 vs. 18.7 mg). The average seed masses of the 30 Grevillea species and 29 Hakea species used in the present study were statistically similar (Table 1). Nutrient assays on the 30 Grevillea species and 29 Hakea species revealed significantly higher concentrations of N and P in the Hakea seeds than in the Grevillea seeds (Table 1; Table S1). The total N and P contents per seed in Hakea were also much higher than in Grevillea (N: 2.66 vs. 0.56 mg, P = 0.002; P: 0.47 vs. 0.17 mg, P = 0.008; Fig. 1).
The seeds of the 25 Hakea species that were tested readily germinated without requiring pre-treatment, and they had an overall germination rate of 82%. Grevillea seeds (29 species) pre-treated with smoke water exhibited a significantly lower germination rate of 32% (Table 1). However, for the 24 Hakea and 12 Grevillea species that yielded a sufficient number of germinants for growth comparisons, the time period to expand the first true leaves was similar, although Grevillea plants had significantly fewer leaves after growing for 100 days. Both Hakea and Grevillea diverted more resources to the roots than to the shoots, with a statistically similar root:shoot ratio. Grevillea and Hakea species accumulated similar amounts of dry biomass and had similar root depths in the soil after 100 days of growth (Table 1). Further analysis of the seed nutrient content and seedling growth rate revealed that Grevillea species had significantly higher nutrient-use efficiencies (Table 1; Fig. 2). Compared with Hakea, for each unit of P in the seed, Grevillea accumulated 1.6-fold more dry biomass and had roots that descended 2.9 times deeper into the soil, whereas for each unit of N in the seed, Grevillea accumulated 1.4-fold more dry biomass and had taproots that descended 3.1 times deeper into the soil (Table 1, Fig. 2).

Discussion
The seed nutrient concentration in Grevillea reported in the present study, particularly with respect to P, was only half of that reported in previous scattered studies. For example, Hocking reported an average of 11.4 ± 2.6 mg.g −1 P for ten Grevillea species 17 . The results of the present study on 30 species reported

Grevillea Hakea
Seed P (mg)  an average of 6.3 ± 2.3 mg.g −1 . This discrepancy is unlikely to be a consequence of methodological errors in our analysis because the 29 Hakea species were assayed in the same batch in the present study, and they yielded an average P content of 13.6 ± 3.8 mg.g −1 , which is similar to the 15.1 ± 4.2 mg.g −1 P concentration reported for 13 species from ten previous studies (except Hakea pycnoneura, which was reported to have an unusually high P concentration of 36 mg.g −1 ) 18 . Interestingly, although the N and P concentrations in the seeds were lower in Grevillea than in Hakea, the N:P ratios in both taxa groups were similar. Notably, the lower seed N and P concentrations in Grevillea relative to those in Hakea did not reflect dilution by seed mass because the tested species in the two groups had similar seed masses.
With only half the seed nutrient content compared with species from the closely related genus Hakea, the species in Grevillea achieved similar levels of seedling growth during the early stages. Indeed, both groups of plants expanded their first true leaves within a similar time, accumulated similar amounts of dry biomass, and diverted similar proportions of resources to the shoots and roots. Consequently, the results of this analysis suggest that compared with Hakea, Grevillea species use nutrients significantly more efficiently than do Hakea. This efficiency represents a clear ecological advantage over Hakea because Grevillea species can accumulate twice as much dry biomass and descend their roots three times deeper into the soil for every unit of nutrient in their seeds.
Seed-stored nutrients (N and P) are crucial for the successful establishment of seedlings in nutrient-impoverished landscapes. The rapid descent of the taproots to reach more reliable soil moisture reserves prior to the onset of summer droughts is critical for the survival of seedlings, particularly in Mediterranean climates with dry summer conditions 19 . Using fewer nutrients, Grevillea species can grow roots equal in length to those of Hakea and survive dry Mediterranean-type summers equally as well, and these characteristics provide a selective advantage to Grevillea compared with other organisms that might require additional nutrients in nutrient-impoverished soils.
Seed production is inversely proportional to the per-seed reserve of nutrients, particularly P, in nutrient-impoverished soils 20 . In the fire-prone environments of SWA, which are generally characterised by poor soil, resprouters (populations that predominantly regenerate through sprouts from the trunk or underground organs) typically produce fewer seeds than nonsprouters (populations that predominantly regenerate through seeds) 21 . However, no significant differences were observed in the P content between nonsprouting and resprouting species growing on P-impoverished soil in a survey of 41 species in SWA 18 . The lower nutrient requirements for seedling growth in Grevillea relative to Hakea, which has much higher nutrient (N and P) concentrations, might allow Grevillea species to produce more seeds regardless of the post-fire regeneration mode. Because of challenges in gauging the size of the soil seedbank, systematic surveys of Grevillea seed production are rare, although Pickup et al. reported a large seedbank for Grevillea rivularis at 193 ± 73 seeds m -2 22 . Because of its capacity to grow to 2.5 m tall and 3 m wide, this plant species has the potential to produce up to 2000 seeds per plant, which is much higher than that of most Hakea species 23 . Levin 24 proposed that propagule-rich lineages are likely to have high speciation rates because the increased number of seeds increases the opportunities for ecological and geographical speciation, and this hypothesis could also be true for Grevillea. Interestingly, high propagule pressure has also been proposed as one of the mechanisms facilitating successful invasion 25 .
Species in the Proteaceae family grow well in the world's most nutrient-impoverished landscapes, particularly P-impoverished soils. Apart from diverse and efficient nutrient-acquisition strategies 26 , high nutrient-use efficiency is likely another key adaptation. Recent studies have observed high photosynthetic P-use efficiency in SWA Proteaceae species 10 . Lambers et al. reported that Proteaceae species such as Hakea species (Grevillea species were not included) from severely P-impoverished soils extensively replace phospholipids with galactolipids and sulpholipids during leaf development to achieve a high photosynthetic P-use efficiency, and proposed the "sulphur-for-phosphorus" hypothesis 16 . Sulpice et al. 27 observed that Proteaceae species grow with very low levels of ribosomes, particularly plastidic ribosomes, at early stages of leaf development. Many of the above studies involved Hakea species, although they rarely included Grevillea species; thus, additional research is required to test whether Grevillea have developed peak nutrient-use efficiency.
It can be speculated that once Grevillea overcome the barriers imposed by the two most limiting factors to its development, low soil nutrient availability and periodic water availability, these plants can invade most habitats. Indeed, species of Grevillea have been observed throughout Australia. High nutrient-use efficiency has interacted with multiple factors, including myrmecochory, a predominantly sexual reproduction strategy, and a high seed output, to allow the Grevillea genus to diversify into a group with 452 species/subspecies within only 11 million years.

Methods
The seeds of Grevillea and Hakea used for the nutrient analyses and germination and seedling growth comparisons were acquired from Nindethana Seed Service (Albany, Western Australia, Australia). Genus-wide seed mass information was extracted from the Seed Information Database 28 . The nutrient analysis was conducted on seeds from 59 species (29 Hakea and 30 Grevillea, Table S1). These species cover the major intra-generic taxonomy sections and provenances; therefore, they represent a wide diversity of the respective genera. For each species, up to 20 g of dry seeds were analysed for N and P concentrations at the ChemCentre (Perth, Western Australia, Australia) using the combustion method for N and inductively coupled plasma spectrometry for P.
A total of 52 species (25 Hakea and 27 Grevillea species, Table S1) were used in the germination trials. Each species had three replicates, and there were up to 30 seeds per replicate (depending on seed size). Each replicate was contained within a 120-mm sterile Petri dish with 2 layers of sterilized Whatman Grade 1 filter paper and sealed with Parafilm. Petri dishes with Hakea seeds were filled with 5 mL autoclaved deionised water, whereas those with Grevillea seeds (because of inherent dormancy) were filled with 5 mL 10% smoke water solution (Regen 2000 Smokemaster); previous studies have suggested that Grevillea seed germination is enhanced by smoke stimulation 29 . To reduce the risk of fungal growth throughout the germination experiment, the smoke water was passed through a 0.20-μ m filter. Germination was conducted in an environmental chamber at a constant 15 °C with a 12 hour light/dark cycle. The germinants (seeds with radicle emergence ≥ 1 mm) were counted and recorded every two days for a period of up to 60 days. Seeds that did not germinate were assessed for viability using a cut test to visually assess the health of the embryo. The Petri dishes were randomised within the environmental chamber in each experiment.
For each of the species germinated in the above-described germination phase (12 Grevillea and 24 Hakea species, with no or minimal germination recorded in some species), ten germinants (or fewer when available) were individually planted in PVC tubes (100 cm in length and 5 cm in diameter) with a substrate of washed white sand (with no nutrients) and placed in a hoop house. The seedlings were watered (20 mL) every 2 to 3 days and grown for 100 days (from late winter to spring 2014) prior to harvesting. The measurements taken at harvest were leaf number, shoot length, root length and fresh and dry shoot and root mass. The dry biomass of the shoot and root was recorded after the samples were oven dried at 70 °C for 48 hours.
The nutrient-use efficiency (N and P) was defined as the amount of biomass produced or root length gained per unit of nutrient (mg) in the seeds. The nutrient-use efficiencies of Hakea and Grevillea were compared. Comparisons of the parameter values between Grevillea and Hakea were analysed with a t-test and implemented in PAST 30 . P < 0.05 was considered to represent statistical significance.