Methods
Soil samples of known area and depth are taken at sample points throughout the study area. Seed- banks exhibit great variation in seed density and species composition over small areas due to the patchiness of species distribution. You must therefore take many samples to achieve an adequate estimate. Dessaint et al. (1996) present a formal approach to designing seedbank sampling. You may want to sample seedbanks over a large area or even to map it. This can be done by positioning sampling points in relation to vegetation types or on a grid system and locating them using the GPS. There is also often great variation in the seedbank over time, reflecting temporal changes in seed production and germination, for instance due to the seasons. You should either take samples onatleasttwooccasionstoassessthisvariationorchoosethesamplingtimecarefullyforinstance after the late summer peak in seed production and before autumn germination in cool temperate zones. If it is large enough the sample can be dug out, using a quadrat to mark the area. Smaller samples can be taken using a sheet-metal cylinder that has one sharpened rim and is pushed into the soil to a certain depth to remove a soil core (Figure 4.3).Thecoreshouldremainintactwhen it is removed from the soil. A removable cross-bar is useful to aid the pushing of the cylinder into the soil and to use to push the core out of the corer. Aquatic sediments can be sampled with cores in the same way, using SCUBA if necessary (McFarland & Rogers 1998).Alternatively,fordeep waterbodies,adredgegrabcanbesettosampletoaspecificsedimentdepth.Youshouldbe able to estimate roughly the area sampled by the grab.
There is no standard core diameter or depth but generally diameter is in the range 2–20 cm with a depth of 5–20 cm. The diameter should reflect the density of the seedbank and the size of seeds. Most studies in most habitats have used diameters of 2–5 cm. The smaller the cores you use the more samples you will have to take in order to sample a reasonable area, but this is usually desirable in order to sample the variation in the seedbank. The core depth should be decided after considering the aim of the study. Viable seeds are strongly concentrated in the top 2–3 cm of the soil and these are the seeds most likely to be recruited naturally into the community. The more deeply buried seeds will give a more complete picture of the seedbank and will also show the seeds available to be recruited following soil disturbance.
Thesoilcoresaretransportedbacktothelaboratoryinbagstoavoidseedlossorcontamination. Here the core can be separated into layers, if you wish, to find the vertical distribution of seeds in the soil. A common division is into layers of depths 0–2 cm, 2–5 cm and >5 cm. The top layer will contain litter and you could scrape this off and analyse the seeds in this separately. In some studies the seeds of the litter layer have been considered not part of the seedbank and discarded. There is no ecological basis for this idea.
The seedbank in the cores can be estimated by germination tests or by counting seeds. Methods involving flotation of seeds on a saline density gradient are inaccurate and difficult and I shall not discuss them.
In germinationtests the sample is air-dried to kill any vegetation and the soil is spread over a seed tray. If you have large quantities of soil or sediment you can concentrate the seeds using the method of terHeerdt et al. (1996).Thisinvolvesbreakingupthesampleandforcingitthrougha coarse-mesh sieve and then through a fine-mesh sieve with a jet of water to remove both coarse and fine material. The material that passes through the coarse sieve but not through the fine is retained. The two mesh sizes are chosen to represent the extremes in seed size in the sample; terHeerdt et al. (1996) used 4 mm and 0.212 mm. The sample can be spread onto a layer of seed compost or sand to decrease the proportion of seeds that remain buried. The seed trays are placed outside, in a glasshouse or in a growth room with simulated natural conditions (daylength, diurnal temperature variation, humidity, etc.) and the soil is kept moist by watering or misting. If the trays are outside you should protect them from herbivores and contamination by airborne seeds. Germination tests for samples from aquatic communities should usually be carried out in submersed seed trays, either in the original water body or under artificial conditions such as in watertanks,perhapsinstandardculturesolution(McFarland&Rogers 1998). The trays should be monitored frequently (e.g. daily) and any seedlings that emerge are iden- tified and removed. As the germination rate slows, monitoring can become more infrequent. The sample should be stirred occasionally to expose all the seeds. Seedlings are identified using the methods described in the ‘Seed traps’ section. It could be useful to maintain a reference collection of seedlings. Monitoring can stop when nothing has emerged for several weeks. An