Developing extruded seed pellets to overcome soil hydrophobicity and seedling emergence barriers

Handlingeditor:DrFlorenciaYannelli Abstract 1. Globally, soil water repellency is a major constraint to plant establishment, restrictingwater infiltration andmoisture retention in the seed zonewhich results in poor germination and seedling emergence. 2. To address this problem within an ecosystem restoration context, we investigated the use of a surfactant in extruded seed pellets to improve native plant recruitment in water-repellent topsoils of two proteaceous woodland species, Banksia menziesii R.Br (glasshouse trial) and Lambertia inermis R.Br (field trial). In this two-part study, we first examined B. menziesii seedling performance in detail under glasshouse conditions for differences in survival between the extruded pelleting formulations after an induced drought at 12weeks. 3.We demonstrated that therewas no difference in seedling emergence amongst control seed and pellet treatments in B. menziesii. Initially, B. menziesii seedlings emerged faster in the control treatment (non-pelleted control seeds) and had greater initial plant growth (leaf and root production), however by Week 12, seedlings generated from pellets were not significantly different from the control seeds and pellets + surfactant had the greatest number of leaf establishment. 4. Survival after drought of B. menziesii seedlings ranged from 14 to 31 days with pellet + surfactant surviving approximately 2.6 days (11.8%) longer than the control seeds. For the second species, L. inermis, seedling emergenceunder field conditionswas approximately 24% greater in seedlings derived from extruded pellets; however, there was no difference in overall survival due to post-emergence predation. 5. This study provides a proof of concept that seedling emergence in water-repellent soils can be enhancedwith extrudedpellets containing surfactants.Our demonstration under in situ and ex situ conditions confirms the prospective use of seed enhancement technologies with future development and field-testing warranted.


INTRODUCTION
Soil water repellency (SWR) can be a major constraint to plant establishment and growth (Ruthrof et al., 2019). As a global phenomenon (Dekker, Ritsema, Oostindie, Moore, & Wesseling, 2009;Doerr, Shakesby, & Walsh, 2000), it can negatively affect the establishment of plants in agricultural systems (DeBano, 2000), particularly during restoration . Soil water repellence leads to decreased water infiltration and moisture retention in the upper soil profile often leading to poor germination and seedling survival (DeBano, 2000;Madsen, Kostka, Inouye, & Zvirzdin, 2012). Identifying, quantifying and mediating SWR is therefore central to improving establishment in agricultural or restoration systems.
Despite considerable research, it remains a problem for which few mitigation technologies or solutions have been developed with success in agricultural or restoration activities (Ruthrof et al., 2019). Extruded pellets that contain soil surfactants is an emerging technology in restoration activities that is being developed to deploy and improve native plant establishment within degraded landscapes including overcoming SWR in North America's sagebrush biome (Davies, Boyd, Madsen, Kerby, & Hulet, 2018;Davies, Madsen, & Hulet, 2017;. It is proposed that with the addition of a surfactant during the production of extruded pellets, and after it is sown in situ, precipitation leaches the surfactant from the extruded pellet into the soil, where is adheres onto soil particles and ameliorates water repellency within the immediate seed microsite . High seedling mortality due to extremely water-repellent soils and drought are key limiting factors of restoration success in Western Australia and other Mediterranean ecosystems (McGhie & Posner, 1980;Padilla & Pugnaire, 2007;Roberts & Carbon, 1972). SWR need only to be temporarily alleviated to ensure successful establishment of plants and then SWR can be utilized to provide natural ecological benefits such as improving soil water conservation (Ruthrof et al., 2019).
This study investigated the incorporation of a non-ionic surfactant into extruded pellets to improve the initial stages of native seedling establishment. We used two proteaceous species (Banksia menziesii R.Br. and Lambertia inermis R.Br.) that occur in highly water repellent, deep sandy soils and are important components of Western Australian restoration. We aimed to improve the hydrological function of the microsite surrounding seeds with a surfactant, by improving soil water infiltration and retention, which will lead to enhanced seed germination and seedling survival. The objective was to ascertain if seedling emergence and early-stage establishment in water-repellent soils was greater from extruded pellets that did or did not contain a surfactant, when compared to directly sown seeds.  Nindethana, 2018). Revegetation with these wild sourced species is an expensive endeavour with low success rates from seeds of 5-7% (Turner et al., 2006); consequently, greenstock production and planting are the preferred method, though this results in a 10-fold higher cost (Rokich, 2016), and still a 7% survival rate after the second summer (Stevens, Rokich, Newton, Barrett, & Dixon, 2016). Therefore, increasing the plant establishment success from seeds is a high priority and investment in seed enhancement development could be a cost-effective approach if proven to be a successful tool. (1) directly sown seeds (control, n = 50); (2) seeded pellets (n = 50); and (3) seeded pellets + surfactant (n = 50).

Extruded pellet production
We mimicked local rainfall events for the Mediterranean ecosystems of the area by taking the average monthly rainfall for the past 5 years over the period of this trial (57 mm August-November; Whiteman Park 009263 meteorological station (B.O.M., 2016)) and applied this amount of water to pots through an overhead irrigation system over 12 weeks. The simulated rain was applied twice weekly over this period and totalled approximately 60 mm per month. After 12 weeks, we applied a drought treatment, no watering for 6 weeks (when ambient temperatures reached up to 37 • C).

Ex situ SWR assessment
To determine the level of SWR in each treatment, three soil sampling harvests occurred post-sowing at 4, 8, and 12 weeks. One additional harvest occurred at 18 and 6 weeks after the induced drought. Six replicate pots from each treatment were selected randomly at each harvest. During each harvest, two soil cores were taken using a small core sampler (25 mm wide and 250 mm long); one adjacent to where the seed or pellet had been sown and one at the pots outer edge(∼ 47 mm way from the seed) ( Figure 1). SWR was measured on the cores at 6 and 12 cm depths (see Figure 1). A third soil core (40 mm wide and 240 mm long) was taken at the centre of the pot, surrounding where the seed or pellet had been sown. This larger core sampler was sectioned into three depths, surface (1-6 cm), middle (6-12 cm) and bottom (12-18 cm). This tested SWR that was (1) inclusive of the seed or pellet zone, (2) the zone immediately beneath the seed/pellet zone and (3) at the base of the pot. Initial SWR was tested using the Water Drop Penetration Time (WDPT) test, using the Dekker et al. (2001) and (2009) protocols for field-moist samples. Three drops of distilled water were placed on the surface of the core soil samples, using a standard medicine dropper, and the time until the drops penetrated into the soil was determined using a timer. Once WDPT measurements were performed on all cores, roots and plant material present were removed and bagged, central large cores were emptied into separate plastic containers and the containers of soil were sealed and weighed.
SWR was assessed at room temperature (26 • C) and after drying the samples at 60 and 105 • C as in Dekker et al. (2001). Soils were dried at 60 • C to replicate the extreme soil temperature that can be experienced in Banksia woodlands during summer (up to 67 • C; Merritt, Turner, Clarke, and Dixon, 2007). After each drying treatment, samples were weighed for gravimetric water content measurements.
We measured SWR of all samples immediately after recording their weight under ambient laboratory conditions (i.e. 20-24 • C / 50% relative humidity). We applied a seven-class index to quantitatively classify the persistence of SWR outlined by Dekker and Jungerius (1990); Class 0, wettable, non-water repellent (infiltration within 5 s), Class 1 (slightly water repellent (5-60 s); Class 2, strongly water repellent (60-600 s); Class 3, severely water repellent (600-3600 s); and extremely water-repellent Classes; 4, (1 hr), 5 (1-3 hr); and 6 (>6 hr). Volumetric water contents were calculated using the gravimetric water content and the bulk density of each top core (0 -6 cm) at Weeks 4 and 8 to calculate the water content present to each seed during the germination and emergence period.

Ex situ seedling emergence and plant growth assessments
Emergence, number of true leaves and survival (after drought) were recorded every 2 days until all plants were harvested. Harvested leaves and roots at 4, 8, 12 and 18 weeks were measured using an EPSON Expression 11000XL photo scanner and analysed using imaging software (WinRhizo software v2007, Regent Instruments Inc. Quebec City, Canada). Leaves were measured for total surface area and total shoot surface area. Roots were partitioned into 10 diameter classes starting at <0.5 mm, to 0.5-1.0 mm, followed by increments of 0.5 mm until reaching >4.5 mm. The software debris removal filter was set to discount objects less than 0.2 cm 2 with a length/width ratio <5. Data for total root length, root surface area, root volume and diameter class length (root length within a diameter class) were generated (in Win-RHIZO) from root images. Shoot and root tissues were then dried in a fan-forced oven at 80˚C for 3 days and then weighed. Percentage emergence was counted on day 57 to replicate the in situ experiment.  reliance of community members to install the trial in the remote field site south of Perth. To minimize any effect at the plot-level, pellets were sown 60 cm apart from each other to act as individual replicates within the broader plot layout. Although we cannot completely rule out the potential of the single site confounding some of our results, we believe that any main effects observed would come from the seed/pellet treatment level.

Statistical analysis
One-way analyses of variance (ANOVAs) were carried out on emergence and measurements of above and belowground growth, average leaf surface area, average root length (cm) and average root surface area (cm 2 ) for the B. menziesii trial. One-way ANOVAs were performed for the volumetric water content sampled from Weeks 4 and 8. Where required, data were log transformed to meet the assumptions for ANOVA analyses. Tukey's honestly significant difference (HSD) test was calculated for multiple comparisons of the mean values for each treatment.
Given the reduced level of replication at the L. inermis trial site, generalized linear models (GLM) fitted with a binomial distribution were used for emergence and survival to use each pellet as a replicate.

Ex situ SWR and soil water content assessment
At the time of sowing, the average WDPT of the soil was over 4 hr, indicating extreme water repellency ( Figure 2). All soils harvested at Weeks 4, 8 and 12 were classified as Class 0 and 1, being wettable (non-water repellent, infiltration within 5 s) to slightly water repellent (5-60 s). After drying at 60˚C, WDPT became more variable between treatments and time to infiltration increased, with Weeks 8 and 12 in Class 1 and 2. Water infiltration times were higher for control seeds (C) in Weeks 8 and 12 had in comparison to the pellet treatments. After soil was completely dry (105˚C), all soil samples were extremely water repellent (Class 5, 3-6 hr), with only pellet + surfactant (PS) indicating lower infiltration times (Class 4, 1-3 hr) (Figure 2). Volumetric water content (%) was significantly greater within soil cores from the seed zone in pots containing a pellet (with and without a surfactant, P and PS) than from pots with control seeds (C) at both Week 4 and 8 (Table 2, Figure 3). No significant differences were found in volumetric water content (%) within soil cores from the root and root elongation zones between treatments (Figure 3). At 18 weeks, and 6 weeks post-drought, all pots were completely dry.

Ex situ and in situ seedling emergence
Pelleted seeds of B. menziesii emerged marginally slower from the soil (time to 25% emergence (T 25 ) was 5.8-11.9% slower in the pelleted treatments P and PS; Table 2), than the non-pelleted seeds (C). By ison to the control seeds (C = 37%) over all treatments (fenced and unfenced) ( Table 2).

Ex situ plant growth assessment
At the first sign of true leaves (45 days), the control B. menziesii seedlings (C) had significantly higher mean number of true leaves than pelleted seedlings (F 2,64 = 8.96, P < 0.001) ( Table 2). At the cessation of leaf production post-drought (90 days), pellets (P) were not significantly different from the control seeds, and pellets + surfactant (PS) had the greatest number of true leaves (Table 2). At Week 8 and 12, the control seeds (C) had greater shoot mass in comparison to the pelleted treatments (P and PS) ( Figure 4); however, by Week 18 there were no differences between treatments. Root biomass did not differ between treatments across the 18 weeks ( Figure 4). No differences were found between treatments for any of the root measurements: diameter classes, volume (data not shown), total length and surface area (Supplementary Information Figure S1).

Ex situ and in situ seedling survival assessment
After imposing a drought treatment in the glasshouse at 18 weeks growth of B. menziesii seedlings, survival ranged from 14 to 31 days (post-drought), with pellet + surfactant surviving approximately 2.6 days longer than the control seeds (Table 2; Figure 5). No seedlings F I G U R E 5 Mean survival (days) post-induced drought at 12 weeks of B. menziesii seedlings grown within different treatments. C: direct sown seeds (control); P: pellet with seeds; PS: pellet with seeds + surfactant (ASET-4001). Significant differences between treatments were identified using post-hoc tests; treatments that do not share a letter showed significant variation (P < 0.05) of L. inermis survived in the unfenced plots due to grazing (kangaroos) and damage by Red-legged earth mites (RLEM, Halotydeus destructor), and of those that survived in the fenced plots (<5%), there was no difference in survival between treatments (control seeds, pellets and pellets + surfactant) (Table 2) due to the RLEM damage.

DISCUSSION
Enhancing microsite conditions to promote seedling emergence is essential to restoration projects, especially considering the predicted drying climatic conditions in Mediterranean ecosystems (Bates, Hope, Ryan, Smith, & Charles, 2008). This study provides a proof of concept that the establishment of woodland species in water-repellent soils can be enhanced using seed pellets. Our demonstration under in situ and ex situ conditions highlights the prospective use of seed enhancement technologies, particularly the use of surfactants and extruded pellets, to overcome limitations to restoration success.
Therefore, any treatments that promote higher or earlier recruitment are likely to benefit many seeding efforts. The observed 20% increase in seedling emergence in L. inermis from pellets used in this study suggests that the immediate seed-soil microsite was enhanced under field conditions. Whether this was due to a break down in water repellency or other in-direct benefits of moisture capture, such as moisture wicking from deeper soils layers , warrants further investigation.
Further, extruded pellets provided a favourable medium for seedling emergence, early-stage establishment and growth in B. menziesii.
Though early life stage demographic processes differed somewhat for un-pelleted B. menziesii seeds (i.e. slightly higher above ground growth in Weeks 4-12), by Week 18, there were no differences in growth performance measured. That is, all seeded and pelleted treatments displayed comparable emergence values and seedling characteristics at the juvenile life stage and some evidence of improved drought tolerance. These findings support our study objective that extruded pellets, and the potential targeted use of soil surfactants, can aid early-stage seedling recruitment.
SWR in this Mediterranean environment is a common occurrence (Ruthrof et al., 2019); it can be severe to extreme (WDPT over 5 hr (Muñoz-Rojas et al., 2016)) and is therefore the primary barrier that we aimed to overcome in this study. Although no distinctive trends indicated that the pellet treatments altered SWR on these soils, there were some specific positive responses. For example, volumetric water content was significantly higher within the seed zone (0-6 cm) of pots containing pellets, indicating the advantageous benefits that pellets, with or without surfactants, have in providing greater moisture during the initial critical growth stages (at Weeks 4 and 8). This finding is con- While seedlings only persisted marginally longer post-drought, the capacity to extend the seedling survival window is a promising result.
In our study, the average survival percentage of L. inermis seedlings was marginally higher in pellets ± surfactant, when compared to un-pelleted seeds under field conditions. Further, seedlings of B.
menziesii growing from pellets had slightly higher survival in drought conditions. Therefore, there is some evidence to suggest that there could be prolonged moisture retention around the root zone promot-  Ruthrof et al., 2019). From this initial pilot study, optimal testing of surfactants (e.g. surfactant concentrations) and other products used to reduce SWR need examination for their efficacy, their incorporation into pellets and their impact on native plant establishment (Müller & Deurer, 2011).
In this study RLEM (H. destructor), an introduced pest of agricultural lands in Australia, extensively damaged the Lambertia field trial seedlings. Additions of insecticides, pesticides and/or systemic insecticides to the pellets could potentially eliminate this next limitation to survival and help the Lambertia seedlings transition to the next life stage (James et al., 2011).
Further investigation into the development of products to help ameliorate SWR at the micro-scale is required (Ruthrof et al., 2019). With greater knowledge of this abiotic barrier arising, there are greater innovative opportunities to refine seed enhancement technologies.
Investigation into the application of seed enhancement technologies to address later-stage seedling dynamics such as to improve seedling establishment and plant survival by overcoming herbivory is also warranted. Our results indicate that pellets may improve the emergence and establishment of native shrub/tree species that are grown in unfavourable hydrophobic environments undergoing restoration. The application of this restoration seeding approach may enhance the establishment of species that do not exist in the soil seed bank or where soil seed banks are not an available resource, by providing a favourable microsite for germination and a longer window of moisture for survival during soil drying.

AUTHORS' CONTRIBUTIONS
AR, JS and TE conceived and designed the research; AR performed the glasshouse and laboratory experiments and analysed the data; AR, JS and TE wrote and edited the manuscript.