Design

Seeds were buried in a minicontainer system developed for litter decomposition studies (Eisenbeis et al. 1995). 75 seeds and 0.5 ml of dried soil from the respective experimental site were filled into small tube-shaped PE-minicontainers (height: 16 mm, Ø: 11 mm). The openings were sealed with gauze differing in mesh size according to three treatments which were prepared to allow access of different organisms: 20 µm mesh size penetrable only by microfauna; 20 µm + F treated with fungicide and ~2 mm accessible to microfauna, mesofauna and small macrofauna. Containers of the latter treatment were sealed with cuttings from a pantyhose with approximately 2 mm mesh size, which could be stretched to enable the passage of organisms with a diameter of up to 4 mm but prevented the escape of seeds. The

1 Only very few organisms were found in the extraction vessels. The second analysis revealed that many small Collembola had desiccated in the soil. I therefore chose extraction by water for further assessments.

containers were then placed into perforated PVC-bars which were pushed horizontally into the soil for burial at a depth of 4 cm (Fig. 5.1). Each bar was equipped with four minicontainers per treatment, later pooled to constitute one replicate of 300 seeds. One bar was buried in each of 6 blocks per site, resulting in six replicates per treatment and site.

Further six bars were placed into site AF3 for early exhumation after six months for a preliminary check-up on seed persistence (data not shown).

Fig. 5.1: Eisenbeis minicontainer bar buried horizontally 4 cm beneath soil level. Minicontainers filled with seeds (with or without fungicide) were sealed with gauze of different mesh size according to treatment (2 mm or 20 µm).

Seed burial was repeated in 2008, on three sites which had shown treatment effects in the first burial and on one site with high Collembola density where mesofauna exclusion (see 5.2.3) had not been successful in the first year. I generally applied the same design and treatments, except for the agricultural field where treatment 20 µm + F was substituted by a new treatment with 500 µm mesh size, penetrable by microfauna and small mesofauna. This was done to single out the faunal fraction responsible for indicated seed predation. Furthermore, the number of replicates was raised to 10-12 (1-2 per block).

5.2.3 Minicontainer preparation & fungicide application

I chose seeds of the winter OSR cultivar Smart, which is found in feral populations in Northern Germany (Dietz-Pfeilstetter et al. 2006) and exhibits a high potential for dormancy (Gruber et al. 2004). Seeds harvested in the previous year (2006 for burial 1 and 2007 for burial 2)² were provided by Syngenta Seeds (Bad Salzuflen, Germany) unimbibed, i.e. not coated with fungicides or other chemicals. Damaged or misshapen were excluded. The empty minicontainers and the gauze were sterilised under UV-light for 2 h prior to the experiment.

2 It was necessary to take different batches of seeds for the two burials as storage can greatly reduce the potential

Soil filled into the minicontainers was sieved to 0.63 mm and defaunated by repeated freezing (24 h at -20 °C / 24 h at room temperature / 48 h at -20 °C). For the second burial, I extended the duration of the freezing and thawing cycles (to 48 h / 72 h / 72 h, respectively), as some containers with 20 µm mesh size retrieved from the first burial were inexplicably inhabited by mesofauna.

I coated seeds of treatment 20 µm + F with fungicide prior to burial. In 2007, seeds were soaked in an aqueous solution of chitosan (6 mg/ml; ChitoPlant, ChiPro GmbH, Bremen, Germany) for 10 s, passed through a sieve and left to air-dry. Chitosan is a natural compound obtained from chitin and is known to inhibit numerous soil- and seed-borne plant-pathogenic fungi (reviewed in Badawy & Rabea 2011). As a registered plant strengthener, chitosan can be applied in organic farming and hence in my agricultural fields. The conventional fungicide captan was used for the seed burial in 2008, as chitosan proved to be ineffective. Captan is a broad spectrum heterocyclic nitrogen fungicide commonly used as seed treatment against seedborne pathogens (Agarwal & Sinclair 1997, Torgeson 1969). It is attributed with a high efficacy against seed-rotting organisms (Neergaard 1979) and can indeed reduce the decay of seeds buried in soil (Mitschunas et al. 2009). Seeds were soaked in a 0.8% aqueous solution of captan (prepared from Merpan 80 WDG with 80% captan w/w; Feinchemie Schwebda GmbH, Eschwege, Germany) for five minutes. Dormancy status of seeds was not influenced by soaking seeds in water for the same period according to a test burial (Appendix II.16).

5.2.4 Burial and exhumation

I needed to induce secondary dormancy in seeds to prevent immediate germination. This can be achieved by incubation in relatively dry soil (Pekrun 1994). Hence, all prepared minicontainers were buried on a site with dry sandy soil near the University of Bremen 3-4 weeks before the experimental burial. The area was protected from rainfall, and minicontainer bars were wrapped in cloth to prevent access of seed predators. Burial on the experimental sites took place on 22-29 September 2007³ (burial 1) and on 13-14 November 2008 (burial 2) before dawn or after sunset. Dormancy of OSR seeds can be broken by light or sudden changes in temperature (Schlink 1994). Therefore, seeds were always exhumed from pre-incubation on the day they were re-buried and transported to their destination wrapped in light-blocking cloth and surrounded by soil. I recovered the minicontainer bars after nine

3 Only 1 to 2 sites could be processed per day in summer due to the fact that seeds needed to be buried in twilight

months of burial, in June 2008 (burial 1) and August 2009 (burial 2). One minicontainer bar could not be retrieved on one site.

5.2.5 Seed viability

Seeds were transferred onto dry soil one day after recovery and stored at 15 °C for up to five days, after which seeds which had recently germinated during burial were counted. Water was added to the soil (see below) and the remaining seeds were tested for viability with a germination test. Defaunated⁴ soil (loamy sand from a derelict site near the Centre of Environmental Research and Technology, University Bremen, Germany) sieved to 2 mm was used as test substrate to avoid spread of fungi observed in previous germination tests on filter paper⁵. All seeds from the respective replicate were spread onto the soil in the test vessel. Test vessels for the first burial were petri dishes (Ø = 9 cm) sealed with parafilm and filled to a height of 3 mm with soil at 80% WHC. Another 3 mm of soil were added after five days to reduce fungal spread in the dishes. As this did not solve the problem, other test vessels were used for the second burial: plastic planting pots (Ø = 7 cm) filled with 179 g of soil and watered every two days to 70% WHC (See Appendix II.17, Fig. II.22, for a summary of differences in test conditions between the first and second burial).

The seeds were incubated in a climate chamber (Sanyo MLR-350H) for seven days at a diurnal temperature cycle which stimulates germination (Thompson & Grime 1983) (12 h of light at 25 °C and 12 h of darkness at 15 °C with a constant humidity of 80%). Temperatures were then changed to 25 °C / 3 °C for the ensuing seven days, as stratification is known to break dormancy in OSR (Gruber et al. 2004), and then set back to 25 °C / 15 °C for the remaining test period of 3-4 weeks. Seedlings and germinated seeds (radicle protruded the testa) were removed and counted every 1-2 days. Soft seeds were considered as dead and were removed, and seeds remaining ungerminated at the end of the experiment were tested for viability with tetrazolium chloride (TZ) according to Duffy et al. (2007) (see Appendix II.10).

No TZ test was done for the first burial, as no intact seeds were retrieved from the soil. Seed persistence was calculated for each replicate as the percentage of viable seeds retrieved out of the number of buried seeds. I could not prevent that, for the first burial, a substantial number of seeds was affected by fungi spreading from adjacent seeds. This rarely seemed to prevent germination, but development of a healthy seedling in the petri dish was unlikely. I noted whether seedlings decayed in the germination test in this way, but counted them as viable

4 See Appendix II.11

seeds in the data analysis as the observed fungi appeared to have spread mostly due to the artificial conditions of the petri dishes.

5.2.6 Data analysis

Data were analysed with SPSS 19.0 (SPSS Inc., Chicago, IL, USA). Proportions were arcsine-square-root-transformed prior to analysis to meet the requirements of parametric analyses. I displayed original values unless otherwise noted. Results are reported as significant at p ≤ 0.05. Replicates were treated as lost when the gauze had loosened, and containers colonised with Collembola despite being sealed with gauze of 20 µm mesh size were omitted from the analysis of treatment effects. For sites AF2 and RH3, this meant a severe reduction in the number of replicates, and results from these sites should be interpreted with caution (burial 2007/2008). I tested for treatment effects with one-way ANOVAs performed for each site, followed by Tukey post-hoc tests. Blocks within the sites were included as a random factor if the block effect was significant. Moderate deviations from normality were tolerated if sample sizes were equal, counting on ANOVA’s robustness (Quinn & Keough 2007). In cases of skewed distributions or heterogenous variances, a Kruskal-Wallis H-test was performed instead.

All intact replicates, regardless of unintended entry of mesofauna, were used to test for differences between sites and for the influence of environmental variables. A one-way ANOVA was conducted to analyse the effect of burial site, pooled over treatment, followed by Tukey post-hoc tests. I further tested in single regression analyses whether the environmental variables WHC, SOM and pH showed an effect on the proportion of seeds remaining viable after the burial of 2007/2008. For this, I calculated the mean proportion of viable seeds per site, averaged over all replicates for all treatments from the respective site. A quadratic regression was performed to relate log-transformed pH with the arcsine-square-root-transformed proportion of viable seeds. The analysis was repeated without site AF3, a site with marsh soil which showed extreme values of pH, WHC and SOM. Regressions performed between seed persistence and WHC or SOM were non-significant, both for linear and quadratic regression models and regardless of whether the outlier site was included.