Department of Botany & Zoology, University of Stellenbosch, Private Bag X1, Matieland, 7602 South Africa

Dr Bruce Anderson



Contact details:

Department of Botany & Zoology
University of Stellenbosch
Private Bag X1
Matieland 7602
South Africa

Tel: +27 21 808-3586
Fax: +27 21 808-2405


Seed Dispersal by Rodents

In the fynbos, two dominant seed dispersal methods exist – plants that are dispersed by wind and plants that are dispersed by ants. However, recently a third dispersal mechanism has been discovered: the spiny mouse, Acomys disperses the large nut-like seeds of Leucadendron sessile (Midgley et al . 2002). It seems as though several fynbos species (possibly 100) have seeds which are not dispersed by wind or ants and have seed characteristics similar to L. sessile and are thus possible candidates for rodent dispersal. This may be particularly so in the Cedarberg area where Acomys is exceedingly common. In support of this, the Cedarberg is particularly rich in plants which seem to have rodent dispersed seeds. Amongst these are the rare Leucadendron concavum , known from a single population and the flagship of the Cedarberg, Widdringtonia cedarbergensis . But the importance of rodent dispersal is also highlighted by the number of numerically dominant species, which display characteristics of rodent dispersal such as Willdenowia incurvata , Leucadendron dubium and Leucadendron pubescens.

How does rodent dispersal work?

Plants which are dispersed by rodents drop their seeds during a narrow window period so that there is a glut of seeds in the environment. Rodents are not able to consume all the seeds, which they find and some like Acomys bury the extra seeds in order to take advantage of these plentiful times. These are then presumably consumed at a later date. Typically seeds are buried singly within five metres of where they were dropped, however we have recorded longer dispersal distances of up to eight metres. To find out this information, we glued brightly coloured pieces of string onto Leucadendron seeds and laid them at the bases of their parent plants. We returned the next day to find that the seeds had been removed. By searching for the bright strings, we were able to ascertain what percentage of seed had been buried and what percentage had been eaten.

Although there is no data available, we hypothesize that mice dig up these buried seeds when food is scarce (we have observed that Acomys easily finds seeds which have been buried by people. Thus we think that they rely on smell to find their food rather than memory). It is likely that seeds are buried within some kind of territory so that other mice are less likely to steal from their stock-piles. Acomys seem to live in small family groups and in captivity, mice from different families are aggressive towards each other.

Effects of burn season

We hypothesise that after the initial release of seeds, the seed-bank becomes smaller as mice eat into their buried stock-piles (See Fig 1). This curve may also be strongly dependent on the availability of alternative food sources throughout the year. As fynbos is a fire-driven system, the timing of fire for rodent dispersed plants is crucial. Fires effectively remove the rodents and allow seeds to germinate. Thus a fire which occurs soon after seed release may be very beneficial to rodent dispersed plants because the seed bank is still large. Conversely, a fire that occurs a long time after seed release is likely to have a negative effect on rodent dispersed plants because the mice would have eaten most of their stock-piles. All this is however speculative as this is a newly discovered phenomenon in the fynbos and long term data needs to be collected in order to prove or disprove these theories.

Fig 1. A hypothetical curve of buried seed bank size after seeds are released.

Is there a rodent dispersal syndrome? What seeds are rodent dispersed and what seeds are eaten without being buried?

Since little is known about rodent dispersal in the Cape , we are still uncertain about which plants are rodent dispersed and which are not. Two main catagories of large seeded plants exist in the fynbos: 1) Those that have wings or hairs and are dispersed by wind after fire (serotinous plants). 2) Those that have elaiosomes which are attractive to ants and are also dispersed by them. A third category also exists 3) Seeds that are released en masse (not after fire). These seeds are large nuts. They have no hairs/wings or elaiosomes and are dropped at the base of the parent plants. Using three plant species ( Leucadendron sessile, Willdenowia incurvata , Leucadendron dubium and Leucadendron pubescens ), that rodents may be very important seed dispersers of this species. However, more data is needed to show how common rodent dispersal is in the fynbos and whether it is possible to recognize rodent dispersed seeds through certain key morphological characters. We hypothesise that rodent seeds should be large, correlated with the presence of a hard seed coat. Firstly seeds need to be large for mice to target them as a sufficiently rich food resource. Secondly, seeds with a large seed coat may be important when there is a glut if seeds as it makes the handling time longer. As a result, seeds with a hard coat may be preferentially buried during times of plenty whereas soft coated seeds may be easily and quickly eaten in situ. We propose placing seeds with different morphological features in the field and recording whether mice eat them or bury them. So far, we have shown that two mouse species ( Acomys subspinosa and Gerbillurus paeba ) are involved in seed burial, with Acomys occurring in rocky areas and Gerbillurus on sandy areas

How interdependent is the mutualism between mice and plants?

The more dependent mutualist partners are on each other, the more vulnerable they may be to extinction. This is due to the fact that mutualists may go locally extinct if their partners go extinct. Thus mutualists are susceptible to their own frailties as well as to those of their partners. This is most important for mutualists, which are obligately dependent on each other and less important for facultatively dependent mutualisms.

Although mice are generalists, feeding on many food substances throughout the year, it has been shown that in certain parts of their range, the Acomys breeding season coincides with times of plentiful nutritious food. For example, in areas with rodent pollinated proteas, filled with nectar, Acomys breeding season coincided with flowering season. Mouse density was also dependent on flower density. Equally important but perhaps harder to measure are the advantages that rodent dispersal confers to gene flow and colonisation. Does mouse dispersal enable plants to colonise new areas? Or is rodent dispersal only for genetic movement within a population? This may have profound effects on speciation patterns and persistence of species in the fynbos.

Alternatively the advantage to plants may be due to the fact that buried seeds are not easily found by other seed predators such as Rhabdomus or parasitic insects. Having large seeds may also be very advantageous in terms of fire survival, drought resistance or K-selection strategies. Since neither ants nor wind can disperse/bury such large seeds, mice may be the only alternative. It is therefore important to ascertain what happens to seeds in the absence of dispersing mice (i.e. by simulating the collapse of the mutualism). This may indicate how dependent plants are on rodent dispersal and burial.

The different hoarding strategies of rodents may also affect how advantageous the mutualism is to plants-for example scatter hoarding may disperse seeds evenly, lowering sib competition but also intra-specific competition with non-sib rivals. Alternatively, cache hoarding may be less efficient at lowering competition between sibs if seed in each cache were collected under the same parent plant. Similarly cache horded seeds do not lower intraspecific competition between sibs as effectively as scatter hording. Questions such as these may be best answered using molecular techniques to determine geneflow within and between populations. Molecular techniques may also determine levels of relatedness in caches. Modelling may also be a useful technique to ascertain relative advantages of different dispersal strategies.

Costs of mutualism

Recent data on mutualisms suggest that mutualisms have both costs and benefits attached to them and at times the benfits of a mutualism may be outweighed by the costs. As a result, at certain times or in certain populations, the outcomes of a so called “mutualism” may be antagonistc. In this system, the costs may be easily evaluated by the amount of seeds which are eaten by the rodents instead of being buried. Fluctuations in the outcomes of mutualisms are thought to create a geographical mosaic of co-evolutionary hotspots and coldspots between associated partners. This is an ideal system to determine whether the net outcome of the relationship between plants and mice fluctuates with time and space. In addition it may be possible to determine which factors cause these fluctuations and where co-evolutionary hotspots and coldspots are most likely to occur.

The effects of thieves

Rodent thieves in the form of Rhabdomus and possibly others may also have important effects on the outcome of mutualism. Such rodents consume seeds without burying them. Rhabdomus is diurnal whereas Acomys and most other Cape rodents are nocturnal. In habitats where only Rhabdomus and Acomys are common, it may therefore be possible to partition seed consumption into two catagories: 1) Seeds consumed by Rhabdomus thieves and 2) Seeds consumed by Acomys mutualists. Thus it may be possible to ascertain how severely thieving affects the mutualism at various localities. The effects of thieving are also likely to affect the geographical mosaic of co-evolution and thus the evolution and persistence of this mutualism.

The importance of rodent dispersal


Many rare and also dominant species may rely on rodent dispersal as a gene flow mechanism, colonisation mechanism or as a mechanism to escape other seed predators. Thus, having a good understanding of all processes involved in rodent dispersal will help in the management policies of fynbos systems. Knowing a plant's regeneration requirements is a key factor in managing species properly and rodent burial/dispersal mechanics are likely to give us insight into how these species are likely to be influenced by different burning strategies. Furthermore, these plants can't be treated as isolated systems because many may be highly dependent on their dispersers and vice versa. The long term persistence and evolution of these relationships is likely to be influenced by factors such as the cost of mutualism and the effects of cheaters in the system.

Experimental procedures

How does Rodent dispersal work?

•  We propose a long term experiment where 1000 seeds are placed at different study sites. Each seed will be uniquely numbered and the study site should be visited on a monthly basis to determine what happens to each seed. The dispersal distance from the original deposition site will be recorded as well as whether seeds were eaten or buried. The fate of each seed will be monitored over the course of a single year (commencing from the time that plants release their seeds). In this way, we hope to simulate what is happening to the seed bank at each study site and show how seed bank size decreases with time after seed release. These data will also show whether there is secondary dispersal after the original dispersal event (i.e. do mice dig up and rebury their seeds). We would like to do this in several different study sites: where Acomys is dominant, where Gerbillurus is dominant and where no burial occurs.

•  On a monthly basis, we would like to place 100 seeds at different study sites. The seeds will be left for a single night before we return to record what percentage of seeds where buried, eaten and what the dispersal distances are. Results will indicate whether burial rates are related to time of year. Although experiment number one will also provide us with this data, mice may have depleted the seed bank to such an extent in the latter parts of the year that it would be impossible to find enough seeds to make conclusions on this aspect of the work.

•  Finally, we would like to saturate certain areas with sunflower seed but leave other areas unsaturated. We would then like to place seeds in these different areas and determine whether burial rates differ in areas where mice have been saturated versus areas where mice are not saturated.

Is there a rodent dispersal syndrome?

•  We would like to test whether seeds which show serotinous or myrmechochorus characteristics are more likely to eaten (than buried) than nut-like seeds. We will place seeds from each ecological/morphological type at several deposition sites and then record seed preferences made by the rodents and also which were most likely to be buried/eaten. In addition we would like to show conclusively that seed burial is common in several cape plant species. We would like to test whether rodent dispersal and burial are important in Widdringtonia cedarbergensis, Leucadendron concavum, Willdenowia incurvata, Leucadendron dubium, Leucadendron pubescens, Ceratocaryum argenteum and Cannomois virgata. In this way, we hope to build a profile of seed characters from which we are able to predict what plants are rodent dispersed by simply looking at the seeds.

Which mice are seed dispersers and do other animals disperse “mouse seeds”?

•  In areas where we find seed burial, we would like to do extensive trapping of rodents using Sherman traps in order to determine rodent biodiversity in different habitats. Rodents will all be released after capture.

•  We would like to make nocturnal observations on seed depots using night vision equipment in order to ascertain which mice are collecting seeds.

•  Finally we have to look at the effects of excluding other seed dispersal vectors (wind, ants and birds) and also examining seed dispersal of each of these vectors individually.

How interdependent is the mutualism between mice and plants?

•  Does the mouse breeding season correlate to the time at which rodent dispersed plants release their seeds? Mice will have to be trapped and it must be determined whether females are pregnant or not.

•  How far do mice disperse plant seeds (answered in research question # 1)?

•  What happens to seeds that are not buried? This would simulate the effect of a mutualism collapse if seed dispersing rodents were for some reason to go locally extinct. This question may be answered by following the fate of seeds where no seed dispersing rodents are present. We can also examine the effects of seed parasites on unburied seeds by placing cages around seeds which allow insects to pass through but not rodents. Or seeds could be glued to stakes in the ground which rodents are unable to climb.

•  Do mouse densities correlate to the number of rodent dispersed seeds in a given area?

Effects of thieves

•  Find study sites with low rodent species diversity (i.e. the burier Acomys and the consumer Rhabdomus ). As Rhabdomus is diurnal and Acomys is nocturnal, we can separate the effects of these mice by placing seeds at different times of day. We may also find sites with just Rhabdomus and just Acomys present. By comparing the fates of seeds in these different habitats, it may be possible to determine what effect thieves have on the system.

 Convergence in seed dispersal times

•  This involves checking when plants of different seed dispersal guilds release their seeds. It is assumed that if there is convergence in seed release times, that sympatric rodent dispersed plants should release their seeds at the same time whereas plants within other dispersal guilds should release their seeds at different times.

Cost of mutualism

•  Here we would like to examine a widespread (False Bay to Namaqualand ) species such as Wildenowia incurvata which is rodent dispersed. This species is very widespread and occurs from False Bay to Namaqualand . We would like to examine the seed burial at a number of populations across its entire distribution range in order to ascertain the predictability of costs and benefits to the plants in this system.


Midgley JJ, Anderson B, Bok A and Flemming T. 2002. Scatter-hoarding of Cape Proteaceae nuts by rodents. Evolutionary Ecology Research 4:623-626.


Developed by