Predation of direct sown seed

Agricultural land vs. woodland sites

Removal of seeds by seed-eating birds and rodents is widely acknowledged as an important limiting factor for direct seeding.

Part of the success on agricultural sites is due to the much lower abundance of seed eating animals on these sites, where agricultural practices such as harvesting and ploughing tend to reduce population build up, compared with woodland habitat with much higher resident populations of mice.

Unfortunately, our work has shown that harvesting operations do not appear to have a similar impact on the abundance of seed eating rodent activity on woodland sites, since seed removal from test caches placed out across clear-fell sites in southern England is rapid: a matter of a few nights.

Who are the predators?

A woodmouse

Numerous studies have shown that the principal agents of predation are small mammals in particular mice and voles.

The most common species encountered in our studies is the wood mouse (Apodemus sylvaticus) - see right, which does not hibernate in winter and needs to feed continuously to survive.  Seeds form an important part of their winter diet and they not only remove seeds for immediate consumption, but also carry encountered seeds to temporary underground larders for later consumption.

The effect of seed mixes

Currently, the accepted wisdom is that direct seeding is likely to fail when seed-eating rodents are present.  However, this is based on experience with relatively few species such as oak, and beech that have often been sown as pure species.  In contrast when direct seeding new native woodlands, seed mixes will contain at least 6, possibly up to 12, different species depending on the woodland type being established.

Studies have shown that rodents given a choice tend to prefer to remove seeds of some species over others (e.g. Kollmann et al. 1998).  So it is highly likely that there will be differences in relative predation risk between the species included in direct-seeding seed mixes.

Such differences might preclude direct sowing high risk species, but still permit use of lower risk species; for example direct seeding of low risk species could be combined with planting seedlings of those at high risk, or protection treatments, such as repellents, might be focussed on the high risk category.

What are the seed preferences?

Seeds and fruits of lowland trees and shrubs arranged in approximate order of preference by rodents.
Top: oak, beech, sycamore, and hazel.
Bottom: wayfaring-tree, wild cherry, dogwood, guilder-rose, field maple, sloe, hawthorn, and ash.


Surprisingly, there are very few studies that have studied seed preferences amongst a wide range of species - most usually concentrate on 2 or 3 species.  We have carried out a series of studies with natural populations of mice to investigate predator preference amongst tree and shrub seeds that are likely to be sown for creating lowland native woodland.  Results so far show that there are distinct differences, but these are by no means absolute and vary somewhat between different experiments.

Aggregated results from several sites using the same seed lots of 12 species show that the most highly preferred species is oak.  Whenever a patch of mixed seeds was encountered, the acorns were always removed first and completely.

The next three species ranked in order of preference are beech, sycamore and hazel; these are usually removed after the acorns, and are often (but not always) removed completely.

At the other end of the preference scale come ash and hawthorn seeds which often remain untouched, or when they are partially removed, they are the last species to go.

Preference for seeds of the remaining species is more variable in terms of the order and extent to which they are removed.

The efect of burying directly sown seed

A small plot sown in early winter with four rows of hazel fruits which have rapidly been encountered and removed. Other species such as ash sown in adjacent plots remained untouched throughout winter.
A replicate of an experiment containing seeds of 12 species buried in bowls of sand. After two nights, only acorns and hazel nuts have been searched out, dug up and removed.

The above results suggest that species producing nuts, and sycamore, are likely to be much more vulnerable to predation than other tree and shrub seeds.  Results from experiments using buried seeds support this prediction.

Burying seed has two effects on seed predation:

  • It can reduce the chances of seed being detected by for example masking smell, but this will depend on factors like burial depth, and the porosity of the soil.
  • It increases the time needed to locate and remove seeds compared with freely-available seed on the surface.

For example, using seeds in buried in sand, only acorns, beech and hazel nuts were removed, the remaining species were unpredated, despite exploratory digging.

On a larger-scale trial using direct sown beech, hazel, field maple, ash, hawthorn and dogwood, excavation of sown seeds was confined principally to hazel and beech (average 70% seed sites dug up), compared with less than 15% for the other species; and these differences were reflected in seedling emergence where only a tiny proportion of sown hazel and beech emerged compared with 30% of the other species.

More work is currently underway to test these differences on a more extensive scale.