de Mendonça, P.G. (2003). The yellow-necked-mouse: a case for conservation in Britain. Presentation to the 2^nd International Conference on Rodent Biology and Management, CSIRO, Canberra, 10-14.2.2003.

---

Philippe Gil de Mendonça

The yellow-necked mouse: a case for conservation in Britain

Past and present distribution and abundance

The yellow-necked mouse, Apodemus flavicollis (Rodentia, Muridae), ranges from the Middle East to Western Europe. However, its subspecies A.f.wintoni (photograph 1) is an endemic restricted to England and Wales exclusively. 

Both fossil evidence ⁽¹⁾ (figure 1) and recent observations recorded in the first half of the 20th century (figure 2) indicate that this subspecies extended as far north as Tyne & Wear and Northumberland and lived there for many centuries. More recent records, however, show that its distribution has shrunk considerably, being now essentially restricted to the Welsh border and some southern counties (figures 3 & 4).

Traditional distribution maps tend to be misleading in that they just show the presence/absence of a species by using dots of the same size, irrespective of the actual density of animals. The current distribution map for the yellow-necked mouse is a perfect example of this limitation. It looks like the national "hot spot" is located in Kent, whereas it is actually located in Gwent. The reason for this is that Kent has an extremely active and efficient Mammal Group with many recorders, whereas there are at most three regular recorders in Gwent. A careful analysis of capture results clearly shows that actual densities are extremely low in Eastern England (where the species occurs, densities are on average 15 times lower than in South Wales, cf. figure 5). Densities in South-Western England (Hampshire, Wiltshire, Gloucestershire, Herefordshire) are somewhat higher than in the East, but sites formerly known as "hot spots" tend to fade away in both relative and absolute densities ^(2,3) (figure 6), and eventually disappear. Last but not least, distribution maps make no distinction between regular occurrence and occasional occurrence, e.g. during a peak density year. 

The English population is extremely small and fragmented whereas the population in South-East Wales, which hosts the sole confirmed current "hot spot" for the whole of Britain, reaches exceptionally high numbers in Gwent, with high densities (70+ ind/ha) comparable to those observed in Switzerland ⁽⁴⁾ or Poland ⁽⁵⁾.

During the peak density period of a good yellow-necked mouse year (Autumn 1998) 168 woods, selected because they were believed to match the requirements of the species, were surveyed by The Mammal Society. Less than half of them yielded yellow-necked mice, most of them only at very low density (figure 7). All beechwoods yielded yellow-necked mice, the highest densities being actually recorded in such woods.

Environmental requirements

Habitat quality is paramount to the maintenance of a high density viable population. In particular, woodland composition and structure are critical for yellow-necked mice. Beech, although not the sole factor determining the presence, abundance, and thus viability of yellow-necked mouse populations, is clearly a key element for this rodent. The highest yellow-necked mouse abundance records in Britain and Switzerland all come from mixed beechwoods. Furthermore, it is striking that the current distribution of A.f.wintoni matches very well the past and current distribution of beech (figure 8).

Interestingly, charcoal analyses from archaeological digs ⁽⁶⁾ indicate not only that beech is indeed native to that part of Britain which currently hosts the sole confirmed "hot spot" for the yellow-necked mouse, but also that forest community structure was there, several millennia ago, very similar to the ones that sustain viable populations of yellow-necked mice in South Wales and parts of Switzerland today. Currently, the best sites for yellow-necked mice are ancient (semi)natural woodlands with many species of trees and bushes producing generous amounts of high calorific value seeds, and providing protective cover (photographs 2 & 3). If one food species fails to mast in any given year, there is still plenty from other sources. Such woods, when dominated by beech and containing 25-33% conifers, are the best habitat for yellow-necked mice. Sweet chestnut and yew are also species favourable to yellow-necked mice, who are extremely fond of their seeds. Since sweet chestnut is a relatively fast growing tree, this species might be used for habitat restoration, e.g. to create connections between fragmented remnants of woodland. However, preservation of currently optimal habitat is paramount and must not be obscured by thoughts of hypothetical restoration plans.

Current genetic diversity

According to the field observations discussed above, it is reasonable to expect the small and fragmented populations from England to be genetically impoverished, whereas the high density (meta)population from Gwent should show a much higher polymorphism, comparable to the Swiss (meta)population previously studied.
Levels of heterozygosity have often been used to infer the history of populations, and in particular the occurrence of bottlenecks. However, sampling error due to the small number of polymorphic loci usually examined results in poor estimates of overall heterozygosity ⁽⁷⁾. Furthermore, heterozygosity is actually unsuitable to detect the occurrence and effects of bottlenecks ^(7,8,9) for selection may favour heterozygosity itself. Homozygous individuals may suffer from reduced survival and/or fitness. Population collapses can even trigger an increase in mean heterozygosity of the population as a whole.
In contrast, allelic diversity is a much more reliable indicator of bottlenecks. Indeed, an extreme bottleneck of two individuals would reduce heterozygosity by 25% only, but would leave a maximum of four alleles at any given locus ^(7,8). Therefore, only allelic diversity is considered here.

Polymorphic microsatellite ⁽¹⁰⁾ loci (Radiograph 1) were used to compare the high density population from Vaud (Switzerland) with the low-density population from East Anglia (England). The number of alleles per locus in the East Anglian population appeared to be greatly reduced in comparison with continental Europe. Because of the small number of samples from East Anglia (n=32), computer modelling was performed in order to determine the impact of such a small sample size on allelic diversity estimates. The likely possibility of differing allelic distributions between populations was taken into account in the model. Despite these highly relaxed conditions, the results of 10,000 Monte-Carlo simulations for each locus indicate that such a dramatic reduction in allelic diversity is extremely unlikely to be the result of sampling alone (p<0.0001). These findings suggest that the East Anglian population is indeed genetically depleted, which is consistent with the hypothesis that increased habitat fragmentation and isolation lead to reduced gene flow, which in turn leads to reduced genetic effective population sizes (far below the census sizes!) and inbreeding. This is a cause of concern, for populations that have passed through a severe bottleneck can suffer from a strongly reduced ability to respond to environmental changes. Furthermore and of more immediate concern, matings between close relatives for a few generations will lead to inbreeding depression, the consequences of which include reduced fecundity and/or survival, and exacerbated sensitivity to environmental stress. Inbred populations have thus a higher probability of extinction than outbred populations ⁽⁹⁾. This situation has been totally ignored by all published studies of British yellow-necked mice so far, thus strongly biasing any interpretation.

Edge of distribution effect, inbreeding depression, or impact of degraded habitat?

As stated above, both the distribution and numbers of yellow-necked mice have diminished in Britain. However, numbers remain highest at the outermost limit of their range (Gwent, instead of East Anglia). Furthermore, body condition (figure 9) of specimens from Gwent is similar to that observed in Switzerland, whereas specimens from East Anglia ⁽⁴⁾ and Surrey ⁽¹¹⁾ are significantly smaller (p<0.0001). This last result is probably a consequence of reduced growth and/or survival due to degraded habitat and/or inbreeding depression, but is certainly not an edge of distribution effect. 

Conservation

Yellow-necked mice depend on a high quality habitat, and thus any destruction, degradation or isolation has a deleterious effect on this specialist species. Restoration attempts may fail for many reasons, and preservation must therefore be preferred instead. Sites to be protected must be selected very carefully. They must not only have the right botanical composition, but also the right structure, and not suffer from disturbances (e.g. forestry management, or impact of the public). Flood risk has to be taken into account as well. Yellow-necked mice living in a wood far from any river or stream can nevertheless be flooded for several weeks, as a consequence of poor drainage and waterlogging after heavy rainfall (photograph 4). Indeed, yellow-necked mice nest in underground burrows where temperature is buffered ⁽¹²⁾ but flood risk high. 

Also, it is often wrongly assumed that small sites are too small to be worth protecting (or even surveying!). This is a lethal mistake! Indeed, some of these small sites actually host very high densities of yellow-necked mice. The highest densities recorded in Switzerland and Wales were observed in woods ranging in size from 0.25 to 5.5 ha. Such small woods tend to be uneconomic to manage and are therefore left undisturbed, thus creating an archipelago of small wooded islands interconnecting farther (and possibly larger) woods scattered in a sea of unsuitable agricultural landscape. It is therefore critical that even very small woods are included in surveys and protection schemes. Failure to do so is likely to lead to further physical and thus genetic isolation which may eventually end-up in an extinction vortex ⁽¹³⁾.