Suffolk s Harvest Mice in Focus

Transcription

1 Report to: PTES and The Chadacre Trust Martha Meek Suffolk Wildlife Trust Suffolk s Harvest Mice in Focus Credit: Terry Longley seeing.org.uk

2 1 Suffolk s Harvest Mice in Focus 4081 pellets were analysed from 226 sites 35% of the sites contained harvest mouse remains Harvest mice were found to make up 1.5% of the barn owl s prey Field visits found harvest mice nests at 85% of all sites where they had been reported in owl pellets from that locality Harvest mouse nests were also found at 74% of sites where they had not been found in owl pellets from that area Nests were found in a range of habitats including arable field margins, rough grassland, riversides and other wetland sites Of the ten different habitat types where nests were found, six were associated with water/wetland It was notable that harvest mice were found to be nesting in wild bird/game cover crops, particularly those containing millet Connectivity in the landscape has more influence than habitat type on the number of nests being found, with very well connected sites having the highest number of nests per site Well connected sites are often associated with estuaries, fens and grazing marshes, where there is extensive areas of suitable habitat Harvest mice are widespread in Suffolk, only completely absent in very intensively farmed landscapes with no tall grass margins Barn owls were only reliable in detecting harvest mice where the pellet sample size was 40 or more If good harvest mouse habitat can be identified, nest searching is the preferable option for a surveyor A predictive model for finding harvest mice nests has been developed based on the findings of this study

3 2 Suffolk s Harvest Mice in Focus Introduction Pg. 3 The history of the harvest mouse in Britain Past studies on harvest mice in East Anglia Pellets as evidence of harvest mice Project Aims Methodology ` Pg. 6 Pellet collection Pellet analysis Site selection Site visits Data collection & analysis Results Pg. 14 Results from pellet analysis Results from site visits Wild bird and game cover crops as harvest mouse habitat Discussion Pg. 27 Using barn owls to find harvest mice Searching for harvest mice nests The harvest mouse and waterside habitat The harvest mouse in a farmed landscape Habitat connectivity is critical Adapting to a modern farmed landscape Managing habitat for harvest mice Conclusion Pg. 36 The future for harvest mice in Suffolk Conservation of the harvest mouse: using what has been learnt for the future Project limitations and variables References Pg. 39 Appendix Pg. 41 Other project outcomes Full data sets for figures Acknowledgements Pellet collection sheet Pellet recording sheet Nest search form Map sample Newsletters General factsheet Farmland factsheet (no page numbers for items in italics)

4 3 Suffolk s Harvest Mice in Focus The history of the harvest mouse in Britain The harvest mouse Micromys minutus has always been associated with lowland arable landscapes, commonly portrayed sitting on an ear of wheat. This image is so common yet evidence now suggests it is far from true, with the harvest mouse more likely to be found in a reedbed than a field of wheat. The species was first described by the naturalist Gilbert White in Hampshire in 1767, From the colour, shape, size and manner of nesting, I make no doubt but that the species is nondescript. They are much smaller and more slender and have more of the squirrel or dormouse colour. They never enter into houses; are carried into ricks and barns with the sheaves; abound in harvest, and build their nests amidst the straws of the corn above the ground, and sometimes in thistles. (White, 1767). Harvest Mice with nest in wheat as illustrated in 1900s Credit: George Abbey, 1909 The abundance of harvest mice at harvest time, as described by White, did not appear to change until the 1950s. Many articles were written in the 1950s about harvest mice overwintering in large numbers in straw stacks both in Hampshire and Oxford (Southwick, 1956; Rowe, 1958; Rowe & Taylor, 1964). Anecdotal evidence from Suffolk farmers who were working in the fields in the 1950s also suggest that it was common to see harvest mice both in the field at harvest and in stacks at threshing time. It was not until the widespread introduction of the combine harvester in the late 1950s, that major changes at harvest time were seen. Whole fields began to be cleared in a matter of hours leaving no refuge for small mammals and straw stacks as winter refuges were lost. The harvest mouse was no longer commonly seen at harvest or threshing time and many believed it to be disappearing from our countryside. This perceived decline and lack of surveying continued until 1973 when the Mammal Society commissioned a detailed survey into the status and distribution of the harvest mouse. Over 1000 individual records, mainly of breeding nests, were gathered. These records were concentrated along the east coast, south of Yorkshire and across the southern counties and the south coast, being more sparsely distributed in the west

5 4 Suffolk s Harvest Mice in Focus country. Records were very sparse in Wales and limited to a few scattered records in Scotland. They were considered to be a common species in all of East and South Britain (Harris, 1979). This survey was repeated in to assess for changes in range, 300 of the same 800 sites were visited but only 30% were found to contain harvest mouse nests (Battersby, 2005). This evidence led to the harvest mouse being classified a Biodiversity Action Plan Species due to the apparent 70% decline. A survey by WildCRU at Oxford University also found a steep decline from 2004 to 2007 in numbers of harvest mice in a lowland arable landscape in the Thames valley (Riorden et al, 2007). Past studies suggest however, that harvest mice have cyclical populations which could explain an apparent decline associated with studies conducted over a shorter time interval (Trout, 1978; Harris, 1979). Past studies on harvest mice in East Anglia In contrast to national studies, those carried out in East Anglia do not appear to find an extreme decline in harvest mouse numbers. A short study in Suffolk in 1978 found the harvest mouse to be widespread, present in all 10km squares within Suffolk and furthermore, in three of those squares they were found to be common (Naunton, 1979). A national study on barn owl pellets, looking at changes in diet between 1974 and 1997 found that overall and in Eastern England harvest mice as a prey item increased over this time period (Love et al, 2000). Studies carried out by Perrow & Jordan and Perrow & Jowitt from concentrated mainly on wetlands in East Anglia where they found the harvest mouse to be abundant. They suggested that wetlands may be providing a stronghold for harvest mice in East Anglia whilst local extinctions may have occurred in arable farmland where habitat has been lost to intensification (Perrow & Jordan, Perrow & Jowitt, 1993; 1995; 2003). Recent informal studies in Essex and Bedfordshire have also found the harvest mouse to be widespread in suitable habitat, especially roadside verges in Essex and reed canary grass in Bedfordshire (Dobson; Woolnough; personal communication). From 1980 to 2008 records of harvest mice in Suffolk were widespread (figure 1) but this was over a very long time period.

6 5 Suffolk s Harvest Mice in Focus Figure 1: Historic records of harvest mice and nests in Suffolk ( - years 2000 to 2008, О- years 1980 to 1999) This map shows all historic records of harvest mice and nests as collected by the Suffolk Biological Records Centre from 1980 to 2008 Pellets as evidence of harvest mice The success of the Suffolk Community Barn Owl Project combined with the lack of knowledge on harvest mice in Suffolk gave the perfect opportunity to study harvest mice using barn owl pellets. Barn owls in Suffolk have steadily been increasing in numbers and spreading to new parts of Suffolk. This range expansion has been helped by the erection of over 1000 new nest boxes combined with the creation of new rough grassland hunting ground, much being arable field margins. As barn owls produce pellets they provide the perfect way to study small mammal communities as the remains of mammal skulls and jaws within them can be easily identified. This expansion in the barn owls range provided the perfect opportunity to collect pellets from across Suffolk and analyse them for the presence of harvest mice. Past work on owl pellets has shown that barn owls do prey on harvest mice regularly, being found in pellets from approximately 50% of sites from eastern counties (Glue, 1975). Overall they tend to make up less than 1% of the barn owls diet, showing that the barn owl does not have a big impact on harvest mouse populations (Glue, 1975). They are also a good species to use in mammal studies as barn owls are habitual and frequent the same roost sites for many years making collection of the pellets an easy task (Glue, 1970). Project Aims The overall aim of this project is to find out if barn owl pellets are an effective way to study harvest mice. This will be backed up by extensive field studies to search for nests. The nest searches will also enable a detailed study of the harvest mouse in the county of Suffolk to be carried out, including distribution and habitat preferences. The practical outcome of the project is to provide a conservation monitoring tool and information on effective habitat management for the harvest mouse which can then be used in other parts of the UK.

7 6 Suffolk s Harvest Mice in Focus Pellet collection Pellets were collected by volunteers monitoring barn owl boxes and nesting sites across Suffolk over the summers of 2009 and Collection was begun in 2009 as this was known to be a year with low field vole numbers. This small mammal is the barn owls preferred prey and they exhibit cyclical changes in population sizes. It was thought that a lack of field voles may force the barn owls to prey on a wider range of species, making harvest mice more likely prey items. As most monitoring visits take place in June, most of the pellets for the project were collected in June although some were collected later and some sites re-visited in autumn. Fresh pellets were collected wherever possible, but in some cases older pellets of unknown age were collected. Twenty or more pellets were collected from each site but in some cases the sample was smaller. For each site details were recorded including the date of collection, six figure grid reference, male or female owl roost, age of pellets, site name, habitat description and site of roost (see appendix for pellet collection sheet). Pellet analysis Pellets were analysed by trained volunteers (a total of 72 current barn owl monitors and 52 new volunteers trained at 5 training days) and all the species were recorded by presence of skulls and jawbones (see appendix for pellet recording sheet). Once a sample was finished the skulls and jaws were bagged and labelled and returned for verification to ensure nothing was missed or incorrectly identified. To ensure volunteers were not leaving anything in the pellet debris, a number of random samples were checked and nothing was found. In feeding experiments with captive barn owls, skulls and jawbones have been found to be very effective at calculating the correct numbers and species of prey items (Buckley & Goldsmith, 1975). Pellet analysis training day

8 7 Suffolk s Harvest Mice in Focus Site selection Sites were split into two categories, those where samples of owl pellets had harvest mouse remains (76 sites) and those where harvest mouse remains were absent (147 sites). Site visits were undertaken on 95 sites. Of these 49 were positive for harvest mice in owl pellets and 46 were negative for harvest mice in owl pellets. This represents approximately 50% of fieldwork in each category, in order to test the theory that barn owls can be used as a tool to locate harvest mice on the ground. The sites were selected to represent a good geographical spread across Suffolk. This method did have its limitations as barn owls are concentrated to the east side of Suffolk with a marked lack of breeding sites across the west of the county. There are also more concentrated populations of barn owls to the north of the county along the Waveney valley (figure 2). This gives the project a geographical bias to these parts of the county. Figure 2: Barn owl boxes in Suffolk Over 1000 barn owl boxes have been erected in Suffolk, all in areas containing habitat suitable for breeding barn owls. Approximately 10% of the boxes are used by breeding barn owls.

9 8 Suffolk s Harvest Mice in Focus Site visits All sites were visited across a two year period from October through to March when harvest mouse nests are at their easiest to find. The boundaries of the land visited were defined by the land ownership and varied in size from gardens to large arable businesses. Once on site, the areas to search were chosen based on their suitability for harvest mice. Suitable habitat included grassland left to grow long for approximately three years including field margins, wet areas of any kind with long vegetation present and wild bird/game cover plots. On sites where more than one habitat was present each suitable habitat type was searched. The time searching each area within a site was recorded along with the time taken to find each nest. All areas searched were marked on a map and a grid reference taken for each nest found. The search time was often limited by the size of the habitat block but where it was not, searching time was for approximately 30 minutes. Details were recorded for each nest including the plant species it was found in and the height above ground (see appendix for nest search form and map sample). Nest searching in long grasses Harvest mouse nest Data collection & analysis To interpret the field data all sites visited for nest searches were allocated an overall site habitat type plus a connectivity score. This enabled the differences in numbers of harvest mice nests at different sites to be looked at in relation to the surrounding habitats and landscape. Each individual nest was then allocated a habitat type within the site to look at specific nesting preferences within the wider landscape. The classifications used are below, including definitions and photographic illustrations:

10 9 Suffolk s Harvest Mice in Focus Overall site habitat type: defined as the site where the search took place and the immediate surroundings as seen. 1. Arable with or without associated field margins 2. Arable with significant amounts of dry grassland that are not field margins 3. Arable with associated river valley 4. Lowland wetland grazing marsh, fenland and estuary

11 10 Suffolk s Harvest Mice in Focus Connectivity score (refers to approximately 1km radius of site): 1. Very well connected includes: Sites near lowland rivers with associated habitat Estuarine sites with associated habitat Extensive fenland/grazing marsh/rough grassland Farms with very well connected, tall margins throughout

12 11 Suffolk s Harvest Mice in Focus 2. Some connectivity includes: Sites near rivers where associated habitat is patchy or not present (normally closer to rivers source) Arable sites with some tall margins 3. Very poorly connected includes: Arable sites with no tall margins and no suitable habitat within sight or connected in any way by suitable habitat Suitable habitat refers to long grass where it is uncut for 2-3+ years or habitat associated with water e.g. reedbeds, ditch vegetation such as reed canary grass etc. It does not include hedges where surrounding vegetation is very short.

13 12 Suffolk s Harvest Mice in Focus Habitat types for individual nests: 1. Arable field margin/corner/ditch (dry or seasonal) 2. Wild bird/game cover crop 3. Rough grassland (where not bordering arable) 4. Tree plantation (where grass is created by plantation) 5. Grazing marsh and grazed dykes

14 13 Suffolk s Harvest Mice in Focus 6. Rivers edge and permanent wet ditch including ungrazed dykes 7. Fen 8. Reed bed Phragmites dominated (wet and dry) 9. Pond edge 10. Estuarine edge

15 14 Suffolk s Harvest Mice in Focus Results from pellet analysis The overall numbers of sites from which pellets were collected was of these sites had pellets collected with harvest mouse remains present and 147 sites had no evidence of harvest mice in the pellets. In total 4007 pellets were analysed from 226 different sites 36% of these sites had evidence of harvest mice in the pellets The number of pellets containing harvest mice was 183 The total number of prey items was 16, of the total prey items were harvest mice The overall percentage of harvest mice as prey items was 1.4% The overall percentage of harvest mice in large samples (20 pellets or more) was 1.5% The average number of prey items per pellet was 4.1 Figure 3: All sites where pellets were collected (red are sites where pellets were positive for harvest Pmice ellet records and blue were negative) 0 positive negative Figure 3 shows all the sites where pellets were collected and the main rivers and tributaries. The percentage of pellet collection sites on the map associated with rivers

16 15 Suffolk s Harvest Mice in Focus is 75% where pellets were positive for harvest mice and 71% where they were negative. The areas pellets were collected from were biased by the collectors some of whom visited more sites than others. The Waveney Valley in the northeast is the area where most pellets were collected, although it is also the area with the highest concentration of breeding barn owls. The west of Suffolk does not support many barn owls so few records came from that area. The east coast was unfortunately underrepresented due to poor pellet collection rates. Table 1: Frequency of Occurence of Harvest Mouse Remains in Pellet Samples - overview and comparison of this study with others Study Location Number of Pellets Sampled Number of Sites Average no. pellets analysed per site Harvest Mice (number) % Harvest Mice as Prey Items Proportion of Sites/Diets where HM found Glue, 1975 Eastern England ca % Buckley & Goldsmith, 1975 Love et al, 2000 East Norfolk ( ) National, 1974 data (range= ) 88% Data not available ibid ibid ibid National, 1997 data Eastern England, 1974 data Eastern England, 1997 data 13, SWT, 2008/11 ibid Suffolk (large samples only) Suffolk (all data) % % Table 1 gives an overview of owl pellet studies, comparing harvest mice in barn owl pellets from 1975 to the current study. Most studies were carried out in East Anglia although some national data is included. There are large variations in the total numbers of pellets sampled, the number of sites they were collected from and the average number of pellets analysed per site. The data does not show large variations according to the location and the year. What it does show is that the probability of finding harvest mice remains in pellets increases with the number of sites and the number of pellets analysed per site.

17 16 Suffolk s Harvest Mice in Focus Comparison of numbers of species found in owl pellets Figure 4: Numbers of species found in all owl pellets The numbers of species shows the field vole to be the most important prey item both in numbers and as prey units (see below). Other species most commonly preyed upon in descending order are the common shrew (26%), wood mouse (12%), pygmy shrew (8%) and bank vole (6%). The harvest mouse makes up only 2% of prey items. Figure 5: Numbers of species in all owl pellets as prey units (calculated by average weight of species as taken from Yalden, 2003) When comparing prey units, some of the larger species such as rat and water vole become more valuable as prey to the barn owl and species such as the pygmy shrew less important due to its small size. The harvest mouse as a prey unit also drops to 1% due to its small size.

18 17 Suffolk s Harvest Mice in Focus Harvest mice relating to small mammal species diversity Figure 6: Number of species recovered in samples of 20 pellets from sites with and without harvest mice present (red-samples with no harvest mice in pellets, blue-samples with harvest mice in pellets) This histogram does show two populations and suggests that the where harvest mice are present small mammal species diversity may be higher. A 2 tailed t-test run on the data is significant but only to a very low p-value (1.72). Minimum owl pellet sample size What is the minimum pellet sample size needed to find harvest mice? The following figures are best estimates based on all the data: pellets would find 50% of all sites with harvest mice pellets would find 85% of all sites with harvest mice The figures are based on mean and median numbers of harvest mice per numbers of pellets of different sample sizes. Confidence limits are based on the assumption that the variation in numbers of harvest mice across sites has a normal distribution. Predictions are based on harvest mice being present at 79% of all sites as found during site visits.

19 18 Suffolk s Harvest Mice in Focus Results from site visits Site visits were carried out from October through until March in the years to sites where owl pellets had been collected. The aim was to search for evidence of harvest mice on the ground by finding the distinctive woven nests. In total 95 different sites were visited, 49 to sites where the pellets collected had been positive for harvest mouse remains and 46 to sites where the pellets were negative for harvest mice. Nests were found at 86% of positive sites and 72% of negative sites. Figure 7: Map of Suffolk showing all 95 site visits carried out - red are visits to positive sites and blue visits to negative sites The sites chosen were based on their geography, although limited by distribution of barn owl boxes which are fewer in the West. The aim was to ensure a good representation of both positive and negative sites across the county of Suffolk.

20 19 Suffolk s Harvest Mice in Focus Figure 8: All harvest mouse records (plotted as 1km2) Figure 8 shows all records for the project from Spring 2009-Spring 2011 plus all the main rivers and tributaries. The records include all breeding nests and non-breeding nests recorded from site visits. They also include some verified records of live mice and dead mice as sent in by members of the public. Records from owl pellets are not included as they can be from a wider area of up to 2km radius, or more in winter, from the pellet collection site. For all the records on the map 73% are associated with the main rivers and tributaries.

21 20 Suffolk s Harvest Mice in Focus Overall percentage of sites where harvest mouse nests were found Figure 9: Percentage of positive pellet sites visited where nests were found Figure 10: Percentage of negative pellet sites visited where nests were found Harvest mouse nests were found at 86% of all sites where the pellets collected were positive for harvest mouse remains and 72% where the pellets were negative. The percentage of sites where nests were found was 14% higher at sites where the pellets collected were positive.

22 21 Suffolk s Harvest Mice in Focus Nests related to pellets and search time Figure 11: Number of harvest mice recovered in samples of 20 or more pellets vs. nests found at 45 sites The majority of samples of pellets (range pellets) contain the remains of between one and five harvest mice and the majority of sites have less then ten nests. The calculated correlation coefficient is which means only 3% of the two variables are in common with each other giving no significant relationship between the number of harvest mice in pellets and the number of nests found. There is still no statistically significant relationship even when all outliers and samples with under 40 pellets are removed (correlation coefficient = 10% of the two variables are in common). Figure 12: Number of nests vs. time to find first nest in 74 areas at 46 sites At the majority of sites it took less than ten minutes to find the first nest and in all sites apart from one, less then 10 nests were found, with most sites having less than five nests. The correlation coefficient is low, (2% in common) giving no statistically significant relationship between search time and numbers of nests found.

23 22 Suffolk s Harvest Mice in Focus Nests found by habitat type Figure 13: Number of nests found in each habitat type The overall numbers of nests found was highest in arable field margin/corner/ditch, wild bird/game cover, rough grassland and rivers edge/wet ditch. The lowest numbers were found in fen, followed by estuarine edge and pond edge. Figure 14: Ratio of nests found in each habitat type to the number of sites where each habitat occurred Figure 14 corrects for the numbers of nests in relation to how many sites were searched containing each habitat type, this then shows which habitats are most likely to contain nests. Wild bird/game cover has the highest numbers of nests, an average of 4.5 per site. Tree plantations also have significantly higher numbers of nests at 2.8 per site whilst all other habitats have an average of only 1-2 nests per site.

24 23 Suffolk s Harvest Mice in Focus Number of harvest mouse nests related to habitat classification 1: Arable 2: Arable with grass 3: Arable with river valley 4: Lowland wetland: grazing marsh, fenland and estuary Figure 15: Percentage of sites where nests were found for different site habitat classifications Nests were found at 65% of habitats classified as arable (1) and arable with grass (2). This figure rose to 85% for arable with river valley (3) and 100% for lowland wetland habitats (4). Figure 16: Nest per site ratio for the different site habitat classifications Once nest numbers had been corrected for the number of sites the habitat containing most nests per site was lowland wetland with 5.25 nests (4). This dropped to 3.75 nests for arable (1), 3.0 nests for arable with river (3) and 2.75 nests for arable with grass (2).

25 24 Suffolk s Harvest Mice in Focus Number of harvest mouse nests related to site connectivity 1: Very well connected 2: Some connectivity 3: Very poorly connected Figure 17: percentage of sites where nests were found for different connectivity scores Nests were found at 100% of very well connected sites (1), 85% of sites with some connectivity (2) and only 40% of sites that were very poorly connected (3). Figure 18: Nest per site ratio for the sites with different connectivity scores Very well connected sites (1) had the highest numbers of nests per site at 5.25, sites with some connectivity (2) had 2.5 nests per site and very poorly connected sites (3) had 1.75 nests per site. (see appendix for full data sets from figures 15 to 18)

26 25 Suffolk s Harvest Mice in Focus Number of nests and connectivity The numbers of nests in arable margins and rough grassland were found to be related to connectivity: (here connectivity refers to the quality of the habitat as there were no other habitat types present e.g. river valleys, to create corridors for movement). In very poorly connected habitats a mean average of one nest was found in arable/grass habitat (based on four sites). In better connected habitats the mean average number of nests in arable/grass rose to 2.75 (based on 24 sites) and in very well connected arable/grass habitats this mean average increased to 8.25 nests (based on four sites). Median scores respectively were 1, 3 & 5.5 nests. The numbers of nests in poorly connected habitat were found to increase where a good food source was present: At the eight sites where harvest mouse nests were found in isolated patches of habitat ranging from 1200m 2 to 13950m 2, there was a marked difference in number of nests depending on habitat type. In the margins and rough grass habitats an average of one nest was found across four sites. In the areas with a good food source present including wild bird/game cover crops and a soft fruit plantation an average of 6.25 nests was found across four sites. Wild bird and game cover crops as harvest mouse habitat Wild bird and game cover crops have been found to support good numbers of harvest mice where they contain the right mixture of plant species. Below is a summary of what was found: In total 15 areas of wild bird/game cover crop were searched Nine out of the 15 wild bird/game cover crops searched contained harvest mouse nests Four sites had harvest mouse nests only in the wild bird/game cover crop Four of the wild bird/game cover crops that did not contain harvest mouse nests contained only maize Two of the wild bird/game cover crops that did not contain harvest mouse nests were mixed plant species including millet but had been stunted by dry weather and were very short One site that had 15 harvest mouse nests in a crop of reed millet and white millet was very poorly connected and had only a small area of rough grass adjacent where no nests were found Eight of the nine wild bird/game cover crops with harvest mouse nests contained millet of some variety although the crops with most nests had at least two varieties of millet Other plant species that were found supporting harvest mouse nests and presumably providing a food source were quinoa and triticale Sunflower and kale were also present in wild bird/game cover crops but were not seen to support harvest mouse nests Most wild bird/game cover crops used maize as a supporting plant species

27 26 Suffolk s Harvest Mice in Focus Figure 19: The average number of nests in different plant species as found in the 15 plots of wild bird/game cover crops Plots containing more than one variety of millet (reed, white and red) contained significantly more nests than plots containing only one variety of millet. Plots with a mix of other species had low numbers of nests and those with only maize had no nests. Nest in game cover crop of reed millet and white millet Close up of nest in a millet spray in a wild bird and game cover crop

28 27 Suffolk s Harvest Mice in Focus Using barn owls to find harvest mice Barn owls have been used as a tool for studying small mammal populations for in many studies, significant for both looking at the diet of the barn owl and the mammals themselves. Past studies have looked at harvest mice as barn owl prey and found them in 50% of diets from southern and eastern counties (Glue, 1975), but have not attempted to back this up with field visits to look at the effectiveness of barn owls as monitors of small mammal populations. The aim of this study was to follow up the pellet analysis with site visits to search for harvest mouse nests, using the results to look for a relationship between the two and the overall effectiveness of the method for surveying the harvest mouse. No trapping was carried out, only nest searches, so it has been assumed throughout the study that there is a positive relationship between the number of nests found and the number of mice. Other studies have found both that there is no relationship between population and numbers of nests (Riorden et al, 2009) and that there is a very good relationship (Kuroe et al, 2011). This contrast is most likely due to the very low numbers of mice trapped in Riordens study which can make results difficult to interpret. In order to test the theory that barn owls are a good tool for finding harvest mice, half the sites visited were where pellets had been negative for harvest mice and half were positive. The theory was not proved as nests were found at 72% of negative sites and 86% of positive sites (figures 9 & 10). The fact that visits to positive sites yielded 14% more sites with harvest mouse nests does show that in some cases the barn owls were effective as monitors of harvest mice. Visits to sites where barn owls were present but there was no suitable habitat for harvest mice are evidence of this. This site is managed for barn owls to encourage field voles but the grass is not long enough to support breeding harvest mice which were not present here or in the owl pellets collected

29 28 Suffolk s Harvest Mice in Focus Overall in Suffolk barn owls and harvest mice were found to have very similar habitats. Although the specific habitat requirements are different the landscapes used by both species are very similar. Small mammal diversity was also found to be higher in pellet samples where harvest mice were present (figure 6) showing that the harvest mouse may be a good indicator species for the health of the habitat and its capacity to support other species. As stated on page 17 the minimum pellet sample size in order to be confident in finding 50% of all sites is 19-20, rising to to be confident in finding 85% of all sites. This means that the method could be used with much more confidence if the sample size could be increased to 40+ for all sites. This does raise questions of the practical application as the time taken to analyse the pellets would increase and in many cases it would be very hard to collect that number of pellets from one site. As the average pellet sample size was only 18 in this study (see table 1), knowing that pellets would find 85% of sites, the proportion of sites positive for harvest mice remains in pellets would have risen if the average sample size was larger. Searching for harvest mouse nests Although nest searching may be thought of a very time consuming process, in reality a good surveyor can assess habitats very quickly and only search those areas that are likely to contain nests. The nests are very distinctive and are rarely confused with other small mammal nests so little training is needed in recognition. In this study ground nests were not recorded as field voles commonly build these type of nests, although other studies have found that harvest mice do build ground nests in winter (Ishiwaka et al, 2010) and also when disturbed as second breeding nests (Trout, 1978). One other type of nest which was not recorded but was found on a few occasions was a tubular shelter often incorporating leaves and always found about 1m off the ground. Ground nest unrecorded in this project Tubular shelter nest unknown species It was thought that there may be a relationship between the time taken to find the first nest and the overall number of nests found. Figure 12 disproves this theory, meaning no assumptions can be made on total numbers of nests even if the first nest is found within a short time. The number of harvest mouse skulls and jaws in the pellets was also found to have no relationship with the number of nests found on the ground (figure 11). This means that numbers of skulls in pellets cannot be used to assess the

30 29 Suffolk s Harvest Mice in Focus size of harvest mice populations. Both these factors make site visits to carry out nest searches invaluable in assessing harvest mouse populations. Nest searches are best carried out from October through until March. Any searches before October will probably result in the disturbance of breeding mice, although it is likely that mice will be breeding until December if the weather is mild. In this study current breeding nests with live young and adult females were seen until November. Nest searches in reed beds and on river banks should be carried out before January as any later than that they will probably have been knocked from the reeds by bad weather or the vegetation may have been flattened by winter floods. Nest searches in grasses can continue until the vegetation starts to grow in April and obscure them from sight. Nests in dry grasses can persist for over a year as demonstrated by a nest monitored from autumn 2009 to autumn 2010 when it was still easily recognised as a harvest mouse nest. Knowing this it should also be taken into account that the previous years nests may be found alongside the current years nests in dry grassland. Harvest mouse nest in dry grassland approximately one year after it was first found in October Although site visits are very valuable owl pellets should not be ruled out as a method for finding harvest mice. In this study half of the sites were pinpointed using owl pellets and the other half the simple presence of barn owls, which would be a good starting point for finding harvest mice in itself as they often share the same habitats. Certainly in a landscape where harvest mice populations were very scattered barn owl pellets may be a very useful tool in finding them. The harvest mouse and waterside habitats The harvest mouse was found to be widely distributed across the county of Suffolk, with nests overall at 79% of all the 95 sites searched. The records are widely distributed across Suffolk with no area having an obvious lack of records (figure 8). 73% of the sites are closely associated with rivers (within 1km), even if the actual habitat they were found in was not riparian. This figure is the same for barn owls based on the pellets collected, suggesting that both species are more likely to be found near a river valley. River valleys are often considered as marginal farming ground as they can be hard to drain so therefore very waterlogged and can also be steeply sloped making

31 30 Suffolk s Harvest Mice in Focus access by modern machinery difficult. For this reason they are often grazed and are sometimes managed for nature conservation creating wildlife habitats and corridors for movement in the process. There is also a legal obligation by farmers to avoid spaying any chemicals close to a water course (including smaller streams and tributaries) which can then lead to an increase in the diversity of plant and invertebrate species. The habitat classification with the greatest likelihood of finding high numbers of harvest mice nests was lowland wetland (figures 15 & 16). Lowland wetlands including fens, grazing marshes and estuaries are more extensive habitats than many river corridors, which are ideal for movement but may not contain enough good habitat for harvest mice to successfully breed. Of the ten habitat types that nests were found in, six are associated with water. Of these habitats the likelihood of finding nests is highest in fens, reed beds and estuarine edges (figure 14). These are the habitats most likely to contain extensive amounts of suitable vegetation for nest building. Grazing marshes where cattle have access to all areas including the ditches support less breeding harvest mice as they often contain little suitable habitat. This grazing marsh has unfenced ditches and high grazing pressure from cattle, meaning that the ditch vegetation is kept short and therefore unsuitable for breeding harvest mice

32 31 Suffolk s Harvest Mice in Focus Pond edges, river edges and wet ditches support less harvest mice overall (figure 14) as they can be isolated and are not always part of a well connected landscape. This is varied though as river edges in this study sometimes provided a narrow corridor only suitable for movement and sometimes very extensive habitat for breeding. Here the deep set river had a narrow strip of rough grassland along its banks, ideal for movement and dispersal to new areas by the harvest mouse. It also supported a very small breeding population, constrained by regular cutting and flooding. The surrounding farmland had a network of tall margins but they were cut too regularly for breeding harvest mice. This river which had a shallower channel provided very extensive habitat because of the way it was managedhere primarily for pheasant shooting but also for wildlife. Wetland vegetation gave way naturally to rough grassland and the area supported a very good population of breeding harvest mice. In contrast to the above site the surrounding farmland was very intensively managed with no tall grass margins. The higher numbers of nests found in wetland habitats, many of which have remained unchanged for centuries, suggests that harvest mice in lowland arable landscapes have always had stable core populations in wetland habitats. Perrow & Jowitt 1995, came to the same conclusion after carrying out detailed studies on harvest mice in the Norfolk Broads. These wetland populations may have prevented the loss of the species from many areas of Suffolk as agriculture intensified over the last century.

33 32 Suffolk s Harvest Mice in Focus The harvest mouse in a farmed landscape The two habitat types which support the smallest numbers of harvest mice are the arable field margins and rough grassland (figures 13 & 14). These are both common habitats but are very varied in quality, easily lost due to cutting and often isolated within farms. The normal margin/rough grass cutting regime for farms within Environmental Stewardship is a three year rotation, although a two year rotation is also widely used. Three years growth is the minimum to support a good population of harvest mice, as only in a few isolated cases were nests found in grasses that had been uncut for two years. As cutting is often carried out in autumn or early winter in the third year of growth, these rough grassland habitats are being removed as they become suitable, often whilst the mice are still breeding. As cutting is rotational the mice will always have somewhere to retreat to but will be unable to build up large populations. It is probable that due to chemical use in the crops that insect numbers are depleted as a food source in most arable field margin habitats. Grass seeds may also be low as a food supply in grass margins, especially if they are are cut before winter when they are needed to sustain overwintering populations of harvest mice, which have very high mortality over winter and early spring (Perrow & Jowitt, 1995; Haberl & Krystufek, 2003). This lack of winter food and loss of suitable breeding habitat at a critical point explains the persistence of only small populations of harvest mice in rough grassland and field margins. The exception to this is the small numbers of farms visited where the management of margins was carried out in such an ideal way for harvest mice that it created a network of habitat throughout the farm, which leads onto the importance of connectivity for the harvest mouse. Here beetle banks which supported good numbers of harvest mice contributed to this farms network of exceptionally high quality margins managed for wildlife conservation. Bence et al, 2003 also found beetle banks to contain significant numbers of harvest mouse nests.

34 33 Suffolk s Harvest Mice in Focus Habitat connectivity is critical Connectivity of habitat is one of the most important factors for the persistence of good populations of breeding harvest mice. The management of margins to increase connectivity within a farm was found to increase the size of harvest mouse populations from an average of one nest on poorly connected farms to eight nests on very well connected farms (see page 25). This demonstrates that not only is the quality of the habitat important but also that increased capacity for movement within sites increases numbers of harvest mice. Connectivity has more effect overall than habitat type on nest numbers, as the number of nests and chance of finding a nest rises steeply with good connectivity and falls rapidly as connectivity decreases (figures 17 & 18). This shows how important it is to reduce fragmentation of habitat which is considered to be one of the major threats to biodiversity. A specific example in this study was a standard arable field margin where a significantly higher than usual number of nests was found. The habitat in the margin was not of a higher quality than usual but was well connected to the Deben estuary by a stream containing reedbeds and fringed by rough grass. This had allowed the mice to migrate from the fringes of the estuary to the field margin. The margin in this picture contained a very high number of harvest mouse nests as it was well connected to the Deben Estuary where harvest mice were breeding successfully in the associated salt marsh and rough grassland. An aerial view of the Deben Estuary showing the location of the margin in the picture above (Image Credit Google Earth, 2010).

35 34 Suffolk s Harvest Mice in Focus A study carried out in Japan by Kuroe et al, 2011, also found that the size of harvest mouse populations is determined by the connectivity of the habitat as well as the quality. It proved that the loss of adjacent habitat lowers population size even if the monitored habitat itself did not change. With the introduction of Environmental Stewardship many farms have reinstated long field margins which have allowed harvest mice to migrate from core riparian populations into the wider arable landscape. It may be the case that harvest mice did undergo a dramatic decline in Suffolk since the 1970s and that in recent years the population has recovered as arable landscapes have become more habitable, although this cannot be proven. Harvest mice are known to be rapid colonisers of new habitat as shown by their preference for newly planted woodlands. After wild bird/game cover, tree plantations were the second most likely habitat type in which to find harvest mice nests (figure 14). A study by Moore et al, 2003, also showed that harvest mice are the most frequently trapped small mammal species in new farm woodland, using the long grasses for nesting. Harvest mice are mobile creatures and although they do not range more than about 15-20m when foraging (Harris, 1979) they are capable of travelling easily over 50m through suitable corridors into new habitats (Perrow & Jowitt, 1995), although this distance may decrease if the corridor through which they are moving is not of good quality (Kuroe et al, 2011). The timing of migrating young is not known but September and October coincides with the movement of mice into winter habitat as studied in Finland, brought about by the onset of the first frosts and halted once cold weather sets in (Harris, 1979). Adapting to a modern farmed landscape Harvest mice are very adaptable species and are generalists in that they are not dependent upon any particular plant species. The most important feature of the vegetation for harvest mice breeding success is the structure. In this study cocksfoot grass, common reed, reed canary grass, reed sweet grass and other upright grasses including false oat grass were the most commonly used plants for nest building. These species are all capable of forming dense cover with a supporting stem structure, both supporting the nest and protecting it from predation and bad weather. The other thing these plants have in common is a wide leaf blade, ideal for splitting and weaving into the nest. The position of the nest varies with the plant species, nests are found low to the ground in dry grasses and over 1m off the ground in reeds and other reed like plants. Another plant which has the correct structure for nest building, plus a good food source is millet, planted in recent years as part of wild bird and game cover plots. Wild bird/game cover is the most likely habitat type in which to find high numbers of nests (figure 14). Millet can be planted very densely and has a very supportive structure with wide leaf blades, similar to reed canary grass which is a common native plant in which to find harvest mouse nests. In the wild bird/game cover crops, the likelihood of finding high numbers of nests increased with the presence of more than one type of millet in the crops. The plot which contained the highest number of nests was a game cover plot for pheasant shooting containing both reed millet and white millet. It was a very poorly connected site with only a small area of rough grassland adjacent and all field margins and hedges were cut very short.

36 35 Suffolk s Harvest Mice in Focus This game cover crop supported high numbers of breeding harvest mice within a very intensively farmed arable landscape It is possible that the mice had persisted in the rough grassland in low numbers allowing them to migrate to the cover plot and build up in numbers over the summer, or it is also possible that they had migrated to the area from a population elsewhere. As the plot would not have been sown until May or reached full height until mid summer and the original number of colonisers was likely to be low, then the population must have grown rapidly. Food supplies from cover plots, which are not ploughed until spring, will also increase winter survival rates in harvest mice, increasing breeding success in summer. In this study millet was just as likely to be planted as part of a game cover plot as a wild bird cover plot, showing that management for shooting as well as farmland bird conservation can have very beneficial impacts on harvest mice populations. Managing habitat for harvest mice As a generalist species harvest mice do not require specialist or complicated habitat management. In general if a good network of habitats, in the form of tall grasses and wetland vegetation can be provided year round and if harvest mice have some way of reaching this habitat, they will probably begin to breed there. The breeding success can then be raised by providing good winter cover and good sources of food, the most obvious source of this being a wild bird/game cover crop including at least two varieties of millet. Obviously management is dependent upon the habitat and this detailed information can be found in the factsheets included in the appendix.

37 36 Suffolk s Harvest Mice in Focus The future for harvest mice in Suffolk Harvest mice are widespread in Suffolk, only completely absent when searched for in very intensively farmed landscapes with no tall grass margins. They have been found in isolated grass habitats in low numbers and in isolated wild bird and game cover crops in high numbers. Where habitat is both of good quality and well connected then they are almost certain to be present. Only at some sites were the nests not found where the habitat was good, these were all in very badly connected landscapes showing the importance of connectivity. Although harvest mice are widespread they are not always common, in many cases only one or two nests were found suggesting that populations are able to persist only at low numbers where conditions are not ideal. In this study over ten nests at one site was seen as exceptionally good numbers but in the 1950s in Britain and in other countries such as Russia, finding 100s of nests or live animals in straw stacks was not uncommon (Trout, 1978). As surveys were not carried out until the 1970s it will never be known how drastically numbers have dropped but it is probable that it is significant due to drastic changes in our farmed landscape. Although numbers may not be as high as they once were the future for harvest mice in Suffolk looks to be secure. They are found on every major river in Suffolk and appear to be moving easily out into surrounding farmland as habitat becomes available. This network of lowland rivers will secure the future of harvest mice even if farming practices in some areas make the landscape inhospitable and with the increase of the use of wild bird and game cover crops harvest mice breeding and overwintering success will increase. Conservation of the harvest mouse: Using what has been learnt for the future This study has demonstrated that the harvest mouse can be found using barn owl pellets but that the presence of barn owls is enough without the pellets to initiate a nest search in the area. If pellets are used then the minimum number to be sure of finding 85% of sites with harvest mice is 40 pellets. This is a high number and very time consuming to analyse, with each pellet taking up to an hour for new volunteers. Where this method may be very useful is if harvest mouse populations are suspected to be very scattered in small clusters and difficult to find on the ground. A good starting point for any harvest mouse nest survey would be to identify areas of good habitat and then visit with a good nest surveyor who would normally be able to find a nest within approximately 10 minutes in good habitat. The likelihood of finding harvest mice nests in the different habitat types is illustrated below as a predictive model. If all the features in any one box can be seen at a site than the likelihood of finding harvest mice nests is very high.