The Yellow-necked Mouse Apodemus flavicollis in Britain: status and analysis of factors affecting distributionAIDAN C. W. MARSH*, SIMON POULTON† AND STEPHEN HARRIS** School
of Biological Sciences, University of Bristol, Woodland Road, Bristol
BS8 1UG ABSTRACT A national
survey of the Yellow-necked Mouse (Apodemus flavicollis) in Britain
was undertaken by The Mammal Society. The live-trapping study sampled
small mammal populations from 168 deciduous woodlands in autumn 1998.
Within their range, Yellow-necked Mice were widespread in deciduous woodland
and were more abundant than Wood Mice in 15% of the woodlands sampled.
These trapping records, as well as records solicited from local recorders,
record centres and individuals, supplemented the existing distribution
map, confirming the general pattern but with minor extensions to some
range borders. Yellow-necked
Mice were found in woodland of all ages but were more common in woods
of ancient origin than in younger woodland. Woodland size was not important
in determining the presence or abundance of Yellow-necked Mice, but they
were more often absent from woods more than 2 km from neighbouring substantial
woodland. The presence
of Yellow-necked Mice did not affect the relative abundance of Wood Mice
(Apodemus sylvaticus). However, the decline in the proportion of
breeding male Wood Mice at the end of the main breeding season was more
marked in those woods that also contained Yellow-necked Mice. Where their
ranges overlapped, Bank Voles (Clethrionomys glareolus) were less
abundant where Yellow-necked Mice were also present. The distribution of the Yellow-necked Mouse was explored with respect to a number of climatic, soil and habitat variables. Maximum summer temperature was the most significant variable explaining distribution, although woodland cover variables also contributed. Soil moisture and pH, mean rainfall and winter temperature parameters did not predict Yellow-necked Mouse distribution. High summer temperature may limit Yellow-necked Mouse distribution through its impact on tree seed production and diversity. Climatic change leading to a rise in summer temperature might encourage range expansion by Yellow-necked Mice, if their other habitat requirements are met. INTRODUCTION In Europe,
the Yellow-necked Mouse (Apodemus flavicollis) is both abundant
and wide-ranging; its distribution extends east into Russia, north into
Scandinavia and south into Greece (Macdonald & Barrett, 1993; Mitchell-Jones
et al., 1999). In central Europe the Yellow-necked Mouse tends to be restricted
to montane areas (Müller, 1972; Saints Girons, 1973) and it is absent
from much of the lowland in the west of the continent. Comprehensive studies
of the Yellow-necked Mouse in Europe have examined allopatric populations
(Pucek et al., 1993; Mazurkiewicz & Rajska-Jurgiel, 1998), as well
as populations living in sympatry with the Wood Mouse (Apodemus sylvaticus)
(Bergstedt, 1965; Hoffmeyer & Hannson, 1974; Gosálbez &
Castién 1995). Despite extensive
data from Europe, there have been relatively few studies of Yellow-necked
Mice in Britain, and these have been confined to limited areas (Hedges,
1966; Yalden, 1971; Corke, 1974; Montgomery, 1977; 1985; Yalden &
Shore, 1991). As a consequence the status and distribution of this species
remains unclear. The most recent distribution map for the Yellow-necked
Mouse (Arnold, 1993) largely relies on casual records and this makes its
accuracy difficult to assess, particularly since Yellow-necked Mice are
so similar in appearance to Wood Mice. The Yellow-necked Mouse has a distinctive
range in Britain, being restricted to the southeast, south and west of
England and central and eastern Wales (Arnold, 1993). The Yellow-necked
Mouse probably became established in Britain in Mesolithic or early Neolithic
times (Yalden, 1982). At this time, it may have been the most abundant
woodland mouse, possibly excluding the Wood Mouse from this habitat (Yalden,
1999). Certainly, in these former times the range of the Yellow-necked
Mouse appears to have been wider, as suggested by archaeological remains
from Roman Manchester (Yalden, 1984), Dowel Cave, Derbyshire and elsewhere
(Yalden, 1999). Older records from the late nineteenth and early twentieth
centuries, where sightings have not been repeated in recent decades, have
led some authors to conclude that numbers and/or range of the Yellow-necked
Mouse may have declined (Morris, 1993; Harris et al., 1995; Macdonald,
Mace & Rushton, 1998). In Britain,
the Yellow-necked Mouse is associated with mature deciduous woodland (Montgomery,
1977; 1985); it is rarely found in other habitats. Woodland with a wide
diversity of seed-bearing tree species, particularly certain key species,
may be important (Yalden & Shore, 1991; Marsh & Harris, 2000a).
The southerly distribution of Domesday woodland and nineteenth century
coppice (Rackham, 1986) show some correlation with the present range of
the Yellow-necked Mouse (Montgomery, 1978), leading to its description
as an ancient woodland species, much as for the Dormouse (Muscardinus
avellanarius) (Bright & Morris, 1990). In coppice woodland the Yellow-necked
Mouse is the only rodent to prefer older, more established compartments,
seldom being recorded in recent coppice (Gurnell, Hicks & Whitbread,
1992; Capizzi & Luiselli, 1996), but little is known about how other
woodland management practices may affect Yellow-necked Mice. However,
populations may decline when woodland is disturbed by extensive felling
or replanting (Montgomery, 1978; Yalden & Shore, 1991). As well as
tree and shrub diversity, other habitat features within woodland are likely
to be important to Yellow-necked Mice. One theory is that Yellow-necked
Mice are more arboreal than Wood Mice, thereby providing a means of spatial
niche separation (Corbet, 1966; Hedges, 1966; Corke, 1974). Hence, the
degree of habitat complexity, the vertical component of woodland habitat
(August, 1983), may be important to their coexistence, although this is
not confirmed (Montgomery, 1980a) The surrounding
landscape, including land use, the distance to neighbouring woodlands,
and connectivity between woodlands, may also affect movement and abundance
in Apodemus (Ylönen, Altner & Stubbe, 1991; Fitzgibbon, 1997;
Tishendorf & Wissel, 1997). Bright (1993) suggested that its presumed
habitat specificity might make the Yellow-necked Mouse vulnerable to habitat
fragmentation. However, is not yet clear that woodland fragmentation does
affect Yellow-necked Mice, or if it does, what degree of woodland isolation
is important. Mader (1984) found that movements of Yellow-necked Mice
were inhibited by the presence of roads dividing forest fragments, but
Kozakiewicz (1993) concluded the opposite. In Poland, Yellow-necked Mice
were found to move across crop fields to neighbouring woods (30-100 m
away) and to disperse up to 300 m over winter (Rajska-Jurgiel, 1992);
they also use crop fields in south-east Britain (D. Corke, pers. comm.).
Since both species of Apodemus can travel more than 1 km in short periods
(Bovet, 1962; Montgomery, 1977; Wolton & Flowerdew, 1985) it seems
unlikely that small distances between woodland fragments would affect
populations. Variations
in climate or soil conditions may also influence the distribution of the
Yellow-necked Mouse in Britain. Corke (1977) suggested its range was limited
to warm, dry areas, while Montgomery (1978) explored the distribution
of wet and dry soil types, concluding that Yellow-necked Mice preferred
drier soils. However, both studies suffered from limited data and analyses
and a more rigorous examination of these and other possible explanatory
variables is required. In this paper,
we present the results from our review of the status and distribution
of the Yellow-necked Mouse in Britain. The first part of this work involved
a national survey of the Yellow-necked Mouse undertaken by The Mammal
Society, the aims of which were: (1) to review and improve the current
distribution map for the Yellow-necked Mouse through the collation of
records from trapping studies and casual recording, (2) to conduct a national
trapping study using skilled volunteers to survey for Yellow-necked Mice
in deciduous woodland across Britain and, (3) to evaluate specific woodland
habitat and landscape variables that may affect the relative abundance
of Yellow-necked Mice. In particular, the following hypotheses were tested:
(i) the range and abundance of the Yellow-necked Mouse is declining in
Britain; (ii) Yellow-necked Mice are more abundant in ancient than in
recent woodland; (iii) Yellow-necked Mice are less likely to be present
in more isolated woodlands; (iv) Yellow-necked Mice are less likely to
be found, or are likely to be less abundant, in woodlands subject to high
levels of human disturbance or management; (v) the abundance of woody
climbers and fallen timber aid vertical movement and are positively correlated
with the abundance of Yellow-necked Mice; and (vi) the structure and abundance
of Wood Mouse populations are not affected by the presence of Yellow-necked
Mice. In the second part of the investigation we analysed the distribution of the Yellow-necked Mouse with the following aims: (4) to review, investigate and interpret the factors that may influence the distribution of the Yellow-necked Mouse in Britain, including climate, soil and woodland area, (5) to explore the potential affect of climate change on the distribution of this species, and (6) to use these data to assess the current conservation status of the Yellow-necked Mouse in Britain. In particular, the following hypotheses were tested: (vii) Yellow-necked Mice are confined to areas experiencing drier soils and lower rainfall; (viii) Yellow-necked Mice are limited to areas subject to warmer summers but are not restricted by cold winters; and (ix) Yellow-necked Mice are associated with higher levels of broadleaved woodland cover.
Methods Study woodlands were chosen by volunteers, but had to be predominantly deciduous and >2 ha in size. Very small woods, pure coniferous woodland and other types of habitat were excluded as these were considered, a priori, to be marginal habitats. Sites were not pre-selected, as this would have reduced involvement, sample size and geographical coverage. Trapping
protocol Traps were
laid down on the afternoon of day 1 and checked as early as possible on
the morning of day 2. They were then reset and checked on the morning
of day 3 before being removed. All animals caught were identified, weighed,
sexed, and their reproductive condition assessed. In order to minimise
any distress to shrews, they were usually identified to species and then
released. Animals captured on day 2 were fur-clipped to identify them
as recaptures if caught again on day 3. Landscape
and habitat variables: selection and scoring In addition, data were collected on planting dates, National Vegetation Classification (NVC) (Rodwell, 1991) and age category. Planting dates were obtained from landowners' records where possible. The age category of each woodland was assessed as either Recent Woodland (RW), Ancient Replanted Woodland (ARW) or Ancient Semi-Natural Woodland (ASNW). ASNW is woodland existing before 1600, ARW is woodland existing before1600 but since replanted and RW is woodland planted post-1600. Data
analysis Missing data
were frequent. Dichotomous coding variables were created to identify unreliable
data and allow them to be excluded from relevant parts of the analysis.
For example, where two or more escapees were recorded on the first trap
round, before the animals were fur-clipped, the "escapes" variable
was marked. Such escapes could have affected the abundance index and this
coding allowed these sites to be excluded from the appropriate parts of
the analysis. The sex of Bank Voles is more difficult to determine than
for mice, so for the purposes of this investigation its sex ratio and
breeding status were not considered. Classifying
rodents as adults, sub-adults and juveniles was subjective and with so
many recorders considerable variation was to be expected. Records of animal
age class were carefully examined; no true difference was found between
sub-adult and juvenile age classes, and so these two categories were merged
into one pre-adult class. To control for inaccurate age classifications
in the resulting two age classes, a reclassification of outliers was implemented
based on weight. All animals that were over two standard deviations outside
the mean weight for their selected age class were reclassified. Reclassifications
were only made once and this lead to the reassignment of 39 (1.5%) Wood
Mice, 7 (1.6%) Yellow-necked Mice and 8 (1.6%) Bank Voles into new age
classes. All continuous
variables were checked for normality and homogeneity of variance. Wood
Mouse abundance was normally distributed after a Log10 (x +1) transformation.
Yellow-necked Mouse and Bank Vole abundance were not normally distributed
and could not be transformed, so non-parametric tests were used on these
data (Siegel & Castellan, 1988). In cases where multiple Chi-squared
or Kruskal-Wallis tests were conducted on these same data, the significant
P values were Bonferroni-corrected; P was divided by the number of tests
conducted to give P' (Altman, 1991). All woodland
and landscape variables (Table 1) were checked for inter-correlation using
Spearman correlation coefficients. LIANA was highly correlated with IVY,
ISO2 with ISO20 and HUMAN with MANAGE. Where variables were highly correlated
(r > 0.7) the variable that contributed least when individually entered
into a logistic regression model was automatically discarded. Logistic
regression was used to produce a model that best explained the differences
observed between woods where Yellow-necked Mice were present and those
where they were absent. Initially the variables were entered individually
into the model and only variables where the Wald statistic had a probability
value of P <0.25 were included in the main logistic regression (Hosmer
& Lemeshow, 1989). In the analysis LIANA, ISO2 and HUMAN were omitted.
The remaining variables were entered simultaneously to produce a predictive
model for Yellow-necked Mouse presence. The trapping date for each wood was recorded and considered in the analysis. Woods were placed in six shorter trapping windows, which represented consecutive fortnights through the survey period. The two woods that were surveyed just before and just after the main survey period were included in the first and last time periods respectively. RESULTS Small
mammal captures Geographical
variation Different variables may influence population density and structure in small mammal populations beyond the range of Yellow-necked Mice, where climatic and landscape differences may be significant. For this reason, sites classed as within or on the border of the Yellow-necked Mouse distribution (henceforth referred to as sites within the range of the Yellow-necked Mouse) were examined separately from those sites outside this range. For sites within the Yellow-necked Mouse range there was a significantly lower abundance of Bank Voles in those woods where Yellow-necked Mice were present than in those were they were not found (Mann-Whitney; U = 1704.5, n = 131, P' <0.05). Abundance of Wood Mice did not differ between these woods (U = 2108.0, n = 134, NS). Differences
between Yellow-necked and Wood Mouse population structure The mean weights of all captures, adults and pre-adults, for the three dominant species are shown in Table 4. The ratio of breeding to non-breeding adult animals was significantly different between the two species of Apodemus ( 2 = 17.88, d.f. = 1, P' <0.001); there was a lower proportion of breeding Wood Mice. However, trapping date had a significant effect on the proportion of breeding animals caught for both Wood Mice ( 2 = 185.8, d.f. = 5, P' <0.001) and, to a much lesser extent, Yellow-necked Mice ( 2 = 14.30, d.f. = 5, P' <0.05). The proportion of breeding Wood Mice appeared to decline rapidly from the second half of October, while Yellow-necked Mice showed a similar decline approximately two weeks later at the beginning of November.
There was no significant difference in the mean weight of Wood Mice in woods where Yellow-necked Mice were or were not recorded. Effects
of landscape and habitat variables The classification
of sites as Recent Woodland (RW), Ancient Replanted Woodland (ARW) or
Ancient Semi-Natural Woodland (ASNW) was provided for 151 of the sites.
Yellow-necked Mice were found in 33% of RW sites, 52% of ARW sites and
53% of ASNW sites. ARW and ASNW sites can be difficult to distinguish,
and so they were combined as "ancient woodland". Yellow-necked
Mice were more likely to be present in ancient woodland than in recent
woodland ( 2 = 4.34, d.f. = 1, P <0.05). The planting date was provided
for 38 sites; four of the seven woods planted between 1900 and 1950 contained
Yellow-necked Mice compared to only one of the seven sites planted since
1950. The vast majority of woodlands (92%) were within 500 m of the nearest woodland larger than 2 ha in size. The distance to the nearest 20+ ha woodland (ISO20) showed a greater variation between sites. Although not stratified for their degree of isolation, all five classes of isolation (<100 m, 101-500 m, 501-1000 m, 1001-2000 m and 2000+ m) were represented by between 25 and 33 sites. Overall, the effect of isolation (ISO20) on Yellow-necked Mouse abundance was not significant (Kruskal-Wallis ANOVA, H = 5.026, d.f. = 4, NS). However, a trend was apparent; a comparison of woodlands >2 km and <2 km from their nearest neighbouring substantial woodland (>20 ha) indicated that Yellow-necked Mice were less likely to be present in the more isolated sites ( 2 = 4.88, d.f. = 1, P <0.05). Correlation
and regression analysis The variables ISO20, MANAGE, IVY, HERB, CANOPY, AGE and NVC were selected for the logistic regression analysis and entered simultaneously into the model. The model (-2LL = 134.374, model 2 = 47.040, d.f. = 11, P <0.001) correctly classified 73% of sites for the presence or absence of Yellow-necked Mice using these variables. The two variables that were found to be significant in this model were MANAGE and IVY (Table 5). The presence of Yellow-necked Mice was positively correlated with the amount of management and inversely correlated with the amount of tree-based Ivy.
Methods Record
collection and reliability Handling
the records Derivation
of variables Data on soil
types in each 1-km square were purchased from the Soil Survey and Land
Research Centre at Cranfield University. These data took the form of a
habitat potential map derived from the national soil mapping programme
1979-1984. The habitat potential map was formed by studying the relationship
between natural and semi-natural habitats and their underlying soils and
amalgamating the associated soil types on the basis of soil wetness, pH
and organic matter. Each 1-km square was assigned to one of 12 categories,
with a few remaining squares uncategorised as "other". We calculated
the total summed value for each of these 12 soil categories for each 10-km
square and these 12 variables were used as independent variables in the
analysis. In addition, the categories were regrouped to create new indices
for soil pH and moisture for each 10-km square. Two further variables describing woodland cover were also used. The areas of both broadleaved and coniferous woodland within each 1-km square were obtained from the Countryside Information System (CIS) (Department of the Environment, Version 5.23) and were originally derived from satellite imaging. The mean of the areas of each woodland type from every 1-km square was used to provide an index of relative woodland cover in each 10-km square. Exploring
the independent variables Before undertaking complex analyses, the 29 variables finally selected (Table 6) were examined conventionally to look for linear relationships between the Yellow-necked Mouse index and the independent variables. Significant relationships were evident for nearly all the independent variables, although most only explained a fairly small part of the variation seen. Summer maximum temperature (23.0%), followed by spring maximum temperature (21.2%) and autumn maximum temperature (16.0%) were responsible for explaining the greatest part of the variation in the index. Matrix plots showed that some weather variables displayed more complex quadratic relationships with the Yellow-necked Mouse index. This indicated that conventional linear statistics would be inadequate to explain these data. Discriminant
Function Analysis (DFA) Analysis
using BEAGLE BEAGLE has
numerous advantages over traditional DFA. Firstly, and probably most importantly
for ecological data, it can uncover non-linear relationships. In particular,
it uses threshold values; for example, "if temperature is greater
than 20°C, then the target is true". Secondly, BEAGLE can use
combinations of categorical, ordinal and ratio scale data in its rules.
Thirdly, it can produce conditional rules using Boolean logic; thus "temperature
> 20oC OR woodland is absent". Finally, because it is a non-parametric
technique it is not adversely influenced by problems such as cross-correlation
in its independent variables. If a rule is bred which utilises two correlated
variables, it is because, at certain values, they are more powerful predictors
in combination than either is alone. Climate
change analysis Results Factors
explaining distribution The overall
rule convincingly predicted the distribution in the east but under-predicted
the range in the west of Effect
of climate change DISCUSSION The
national trapping study The trapping
study marginally extended the known range of the Yellow-necked Mouse into
Staffordshire and Bedfordshire, and records from parts of Cambridgeshire,
north-east Kent and Shropshire clarified more speculative parts of the
distribution map. There were also many new records within the known range,
and the high number of such sites indicates that the Yellow-necked Mouse
is still widespread within suitable woodland inside its existing range.
Previous
studies have shown that Yellow-necked Mice may exist in greater numbers
than Wood Mice (Yalden, 1971; Montgomery, 1980b; Marsh & Harris, 2000a),
and it has generally been assumed that these woodlands represented exceptional
sites or events. In this survey Yellow-necked Mice had a higher relative
abundance than Wood Mice at 15% of sympatric sites, suggesting that within
suitable woodland, at least during the autumn, Yellow-necked Mice are
the dominant species more often than was previously thought. Montgomery
(1980b) showed that loss of breeding status in sympatric Yellow-necked
and Wood Mice occurred much as suggested by this study, with males retaining
breeding status for longer than females. However, it is the structure
of Wood Mouse populations in woods with and without Yellow-necked Mice
that is of particular importance to understanding possible consequences
of sympatric living. Competitive interactions may occur between these
two species (Montgomery, 1981; 1985), although how these manifest remains
uncertain. One suggestion has been that Yellow-necked Mice may be more
arboreal than Wood mice, although efforts to investigate this have been
largely inconclusive (Montgomery, 1980a; Marsh, 1999). Based on removal
experiments, Montgomery (1981) concluded that competitive interactions
between Yellow-necked and Wood Mice are weak and probably ameliorated
by differences in their biology. The results from this study imply that
the presence of Yellow-necked Mice within a wood somehow accelerates the
loss of breeding condition of breeding male Wood Mice, although reasons
for this are not clear. Woodland
age may be important to Yellow-necked Mice, with mature deciduous woodland
the favoured habitat of this species (Montgomery, 1977; 1985). These data
provided one of the first opportunities to explore these ideas on a wider
scale. Although ancient woodland was significantly more likely to contain
Yellow-necked Mice, even woodland planted in the last 100 years could
support Yellow-necked Mice in autumn, although whether mice would be present
at periods of lower abundance is not known. Woodland less than 50 years
old appeared to be less suitable for Yellow-necked Mice. Overall, these
results and those of other studies (Marsh et al., 1999; Marsh & Harris,
2000a), suggest that while ancient woodland may offer the best habitat
for Yellow-necked Mice, mature diverse woodland that does not date back
to pre-1600 may be just as suitable for this species. Woodland
size (>2 ha) was not important in predicting the presence or absence
of Yellow-necked Mice. However, woodlands isolated by more than 2 km from
the next substantial woodland block (>20 ha) were significantly less
likely to contain populations of Yellow-necked Mice. Other studies have
shown that lower isolation distances between woodlands do not appear to
have any discernible effect on the relative abundance of Yellow-necked
Mice (Marsh et al., 1999; Marsh & Harris, 2000a). This supports the
hypothesis that woods need to be well isolated before fragmentation affects
the distribution of this species. However, the age and composition of
neighbouring woodlands might also be important, as might differences in
the intervening habitat type and the connectivity of fragments. We still
know relatively little about dispersal by Yellow-necked Mice and how these
variables may affect such movements. Yellow-necked
Mice were more likely to be present in woods with less Ivy cover on the
trees and higher levels of management activity. These results were at
odds with initial hypotheses that Yellow-necked Mice would prefer sites
showing least disturbance and greater Ivy cover to promote arboreal movement.
However, Ivy is the only British member of the tropical family Araliaceae
and its growth is strongly temperature dependent, displaying late flowering
and great sensitivity to frost (Godwin, 1975). As a result, stems do not
grow up into trees as frequently in the north and east of the country,
where winters are cold and dry, as in the west, where winters are warmer
and wetter. The regional variation in tree-based Ivy scores from this
study reflected this pattern. The European range of the Yellow-necked
Mouse (Mitchell-Jones et al., 1999) correlates with areas that experience
drier, colder winters and we suggest that the relationship between this
species and woods with lower levels of Ivy cover indicates a climatic
effect. We suggest two possible explanations for the ability of higher management scores to help predict the presence of Yellow-necked Mice. Firstly, many of the woods surveyed were nature reserves; these are likely to be high quality woodlands that are often actively managed. Therefore, management activity may have been subtly linked to woodland quality, something that might explain its relevance to the presence of Yellow-necked Mice. Secondly, a high woodland management score did not necessarily indicate the type of disturbance that would be detrimental to Yellow-necked Mice or other rodents; recent evidence of thinning, ride widening or coppicing would have resulted in a high score. When carried out sensitively, these practices may improve the variety of microhabitats within the wood, enhancing the range of potential food sources for rodents. Distribution
analysis The smoothed
distribution (Figure 4) of the Yellow-necked Mouse highlights core and
peripheral areas. The analysis using BEAGLE generated rules that were
good at determining the range of the species in the eastern zone, but
notably less successful for the west of the country. Since the western
part of their range displays more mountainous and varied topography than
the east, one possibility is that the 10-km scale used here was too broad
to detect the microclimatic differences in these areas, particularly across
into Carmarthenshire. Alternatively, some other variable that we have
not included may also be of particular importance in determining their
range in this area. For example, these analyses were unable to take account
of the age of woodland and only differentiated broadleaved and coniferous
woodland. A computerised ancient woodland inventory was not available
but this would undoubtedly have been beneficial to these analyses. Nonetheless,
the predictive capabilities of the rules produced by BEAGLE were impressive.
The repeated selection of summer maximum temperature in these rules and
in the DFA suggests that this is an important variable. Both analyses
suggest that winter temperature is unimportant in defining Yellow-necked
Mouse range. The failure of rainfall, soil moisture, soil pH and most
soil variables to help predict the range of the Yellow-necked Mouse indicates
that these factors may not be important either. However, maximum summer
temperature may be closely related to summer soil moisture deficit, and
so the conclusions of Montgomery (1978), who suggested that Yellow-necked
Mice were limited to areas with drier soils, cannot be entirely discounted.
Nonetheless, high summer temperature was found to be of overriding importance
and we suggest that the most likely explanation may relate to botanical
associations. High summer temperatures are known to affect tree seed production.
If the yellow-necked mouse specialises on tree seed (Corbet & Harris,
1991), then the availability and reliability of seed production is likely
to be particularly important. The laying down of flower buds in Beech
and Oak, as well as other tree species, is favoured by high summer temperatures
(see Matthews, 1955; Jones, 1959) and summer temperature can also influence
tree seed fertility and thereby species distribution (see Pigott &
Huntley, 1978). We know that
Beech mast, acorns and hazelnuts are regularly stored and utilised by
Yellow-necked Mice, so the distributions of these tree species may be
particularly important. Natural Beech woods are restricted to parts of
southern Britain, while native Beech does not reach the far south-western
counties of Somerset, Devon and Cornwall (Watt & Tansley, 1932). In
addition, Quercus robur is the usual cover over Hazel in coppice-with-standard
woodland, but it is rarer than Q. petraea in Wales, Devon and Cornwall
(Jones, 1959). Where Yellow-necked Mice are found in areas not predicted
by our models, such as across South Wales into Carmarthenshire, Quercus
woodland is present against expectations based on climate. These observations
are speculative and there is no evidence to show that any specific association
exists with any tree species. However, it is known that all tree species
show an irregular pattern of annual seed production. Thus a seed specialist
could be restricted to areas where sufficient species diversity occurs
to ensure an adequate food supply each year from one source or another.
Jones (1959) points out that even in years when a region suffers from
almost complete seed failure, there will be abundant seed over a limited
area, a single wood, or a part thereof. This would seem to fit with the
tentative conclusions of Montgomery (1985) and Marsh & Harris (2000a),
who suggested that Yellow-necked Mice might retreat to key woodland areas
over the winter, where tree seed availability remained good. Since the rules produced by BEAGLE found that Yellow-necked Mouse distribution was most heavily influenced by maximum summer temperature, we used this to explore the way in which distribution may be affected by climate change. However, the predicted temperatures used in the model, based on climate change scenarios, were for mean summer temperature, while the model itself was based on maximum summer temperature. These two variables are closely correlated and since the analysis of distribution in relation to predicted climate change was for illustrative purposes only, this discrepancy was not considered crucial. The overall trend will not be affected by this anomaly; the predicted distributions will simply underestimate range expansion, as can been seen in the 2020 low scenario, where predicted range is actually more restricted than the model predicts for the present. The predictions do show that as climate warms Yellow-necked Mice may benefit by expanding their range to the west and the north. Whether such range expansion actually occurs will inevitably depend upon a number of other local factors not highlighted in this broad model. These include the availability of suitable woodland and the accessibility and connectivity of woodland fragments. Summary The work
presented here provides a review of the current status and distribution
of the Yellow-necked Mouse in Britain. However, there are still a number
of areas of Yellow-necked Mouse ecology that would benefit from further
study, not least the comparative diet and arboreality of Yellow-necked
and Wood Mice, details of Yellow-necked Mouse social structure and dispersal
and the impact of tree seed production on their abundance and distribution.
|
