Independent Audit of Badger Trial: Audit of the Surveying and Social Group Delineation SOPs
3.0 Assessing the effectiveness of the "Delineation of badger social group territories" SOP
3.1 Methodology
3.1.1 Field Surveys
A total of four clusters of putative badger social groups was chosen for bait-marking investigations as part of this element of the Audit. Two of these clusters were in triplet B and two in triplet A. The clusters were chosen in order to include areas with a range of BRO histories, and to include territories which appeared to be relatively stable, with well-defined boundaries, as well as those where badgers' likely social organisation was less clear-cut. In addition to these areas, the need for a series of more limited bait-marking investigations was identified for a cluster of setts in triplet B. The scale of these investigations was such that it was sensible to extend the operation and to include this area as a fifth bait-marking cluster.
An explanation of the theory and application of bait-marking techniques is given in Appendix I . In this study, marked bait was fed at each of the main setts (or possible main setts) within the clusters during April and May. Markers were then recovered from latrines and dung-pits throughout the survey areas and surrounding land. Because of the limited number of groups involved in each of the clusters, and the fact that the work was undertaken relatively late in the bait-marking season, an extended period of baiting was used, and care was taken to identify as many latrines and dung-pits as possible.
In addition, since the bait-marking commenced at a relatively late stage in the spring peak in marking behaviour, and several 'gaps' were identified in the apparent territory boundaries, some of the bait-marking investigations were repeated during September and October in order further to refine the results and to help resolve the situations where the badgers' social organisation remained unclear.
3.1.2 Assessment methodologies
For those clusters of setts where detailed bait-marking investigations took place, the bait-marking 'returns' have been mapped according to standard conventions (see Figures 1-5). In addition, an inferred territory boundary has been added with two levels of reliability. Solid lines indicate that a relatively clear territory boundary could be delineated, by a combination of the identification of marked faeces in dung pits or latrines and the recognition of other field signs typical of territory boundaries (e.g. boundary paths and linear aggregations of dung pits and latrines). Dashed lines indicate parts of the territory boundaries where their precise location was unclear because of sporadic marker 'returns' and the general absence of clear boundary signs. Generally, where a territory extends into an area not surveyed as part of the bait-marking investigation, no territory boundary is inferred.
3.1.3 Consultations
Similar discussions took place with regard to this element of the operational procedures as those outlined in section 2.1.3.
3.2 Results and discussion
3.2.1 Interpretation of bait-marking results
Figures 1-5 indicate the results of the bait-marking investigations. Clearly, there are significant variations in the accuracy with which territorial boundaries could be delineated on the basis of the bait-marking 'returns'.
The relative absence of badger activity in the central part of Figure 1 made territory delineation in this area extremely difficult. Similarly, a proportion of the boundaries on Figure 3 (particularly between Groups 2, 3, 4 and 6) were unclear. It is likely that this is a reflection of the effects of previous BROs, both creating areas of 'dead space' and reducing population density, which may, in turn, affect the territoriality and marking behaviour of the remaining animals. The bait-marking 'returns' for Group 3 on Figure 3 are the best example of extremely limited territorial marking behaviour, but the territories of several other groups were characterised by the presence of relatively few large 'typical' boundary latrines and other features characteristic of stable, well-defended territories. Nevertheless, as all the figures show, reasonably distinct territories could be identified, with at least some boundaries being regularly marked and relatively stable (between successive bait-marking periods). On the basis of the bait-marking investigations, and the general impression from the field surveys as a whole, it appeared that each of the study areas contained relatively distinct social groups behaving in a territorial manner: in these areas, the past BROs did not appear to have resulted in a completely fluid, aterritorial system, where the reliable identification of main setts, social groups and (some) territory boundaries was not possible.
For some of the figures, further explanation of the bait-marking results would be helpful:
Figure 1:
The two large setts in the south-eastern corner of the study area were
confirmed as 'belonging' to the same social group. Bait-markers were readily
transferred between them (in both directions) and there was 'cross-marking'
(i.e. markers fed at one sett appearing on or close to the territory boundary
beyond a neighbouring sett) of boundary latrines.
Figure 3:
The large setts in the north-western corner of the study area were confirmed
as 'belonging' to two distinctly separate social groups. A stable, well-maintained
boundary was identified between them, and only a small number of bait-markers
were transferred (on a single occasion) from bait fed at Sett 8 to a latrine
close to one of the setts indicated as Sett '7'. It is assumed that this
represents an occasional territorial transgression, possibly by a male
in search of oestrus females.
Figure 4:
The degree of territory partitioning between Groups 1 and 2 was unclear.
There was transfer, in both directions, of small numbers of bait-markers
between 'main' setts, and a small amount of 'cross-marking'. However,
there was also some indication of a boundary between the two. This pattern
of bait-marking 'returns' is not uncommon, and it is likely that these
are two 'related' groups, with the ranges of some animals encompassing
both territories, whilst other, possibly younger, animals behave as separate
social groups. Relatively clear territorial boundaries were identified
to the west of Group 1, to the north-east of Group 6 and, to some extent,
to the north of Group 7, with several latrines found which did not contain
markers. This pattern of marking suggests the presence of an undetected
group between Groups 1 and 6 however access could not be obtained to this
area during any of the surveys.
Figure 5:
The degree of 'cross-marking' and exchange of markers between the setts
themselves (markers fed at Sett 7 appeared in the latrine close to Sett
8 on several occasions) strongly suggested that Setts 7 and 8 'belong'
to the same social group.
Very small numbers of markers fed at Sett 10 appeared (on one occasion each) in latrines close to Setts 9 and 11. However, it appeared clear that Groups 9, 10 and 11 were separate and that, as with Groups 7 and 8 on Figure 3, this transfer of markers represented occasional territorial transgressions.
3.2.2 Accuracy of tessellated boundaries
Aside from the boundaries shown on Figure 2, there are clearly huge discrepancies between the tessellated boundaries and those derived from the bait-marking investigations. These obvious discrepancies mean that a detailed analysis of the differences (in terms of coincident areas, boundaries or sett allocation) has been difficult. The vast majority of these discrepancies appear to be a direct result of the differences in main sett classification that have been identified in Section 2.2. The only other clear reason for the discrepancies between the tessellated and bait-marking-derived boundaries is illustrated on Figure 4, where a motorway is assumed to be a 'natural' boundary between a succession of groups, when in fact the bait-marking 'returns' indicate that the territories of at least two groups encompass areas on both sides of it.
In the case of the cluster of social groups comprising the triplet B bait-marking investigation (Figure 2), the classification of main setts on the basis of the re-survey was the same as for MAFF's original field survey. Here, there is a general agreement between the tessellated boundaries and those derived from the bait-marking data. There is a significant discrepancy in the case of Group 1, where the relatively clear territory boundary that was identified to the north-west of the main sett falls well within the tessellation boundary (this may indicate the presence of a neighbouring group in this direction, the main sett of which was outside this bait-marking study area and was not identified during MAFF's original survey). However, with the exception of this group, the tessellation- and bait-marking-derived territories are similar. A total of approximately 70% of the bait-marking boundaries coincide with the tessellated lines (at least approximately so), compared to 30% which clearly diverge. Given the limited extent of this particular bait-marking cluster, it is not possible to compare coincident areas. However, possibly the most informative measure is the overall proportion of setts which fall within the 'correct' territory, in this case 100%, although any setts lying to the north-west of Group 1 (which would not have been included in the survey of this cluster) could be incorrectly ascribed to this Group, given the extension of the tessellated territory boundary in this direction. If Group 1 is removed from the calculations, almost all of the boundaries are also roughly coincident.
Although for the other bait-marked 'clusters' there were significant discrepancies between the various boundaries, some further analysis was warranted in order to investigate the potential discrepancies in terms of assigning setts to the different tessellated areas ('tiles'), since this represents the most significant practical application of the tessellation technique.
Before presenting these data, it would be appropriate to review the various difficulties encountered in attempting to make definitive assessments as to the errors in assigning setts caused by discrepancies in the tessellated boundaries:
(a) There were limitations in the extent of bait-marking (it was necessary to stop somewhere), which meant that not all of the circumference of each territory boundary was delineated for all the groups in a 'cluster'. Thus there was not always a full complement of assigned setts for each group territory/tile. In addition, some territories appear to be missing as a result of access restrictions (e.g. on Figure 4).
(b) On several occasions the tessellated boundaries differ so comprehensively from the baitmarking-inferred boundaries, it was not clear which to choose as the 'correct' tile/group territory (e.g. Group 10 on Figure 1; Group 7 on Figure 4; and Group 5 on Figure 5).
(c) How to deal with tiles which more or less encompass two groups (e.g. Groups 2 and 3 on Figure 1 and Groups 5 and 6 on Figure 3).
(d) None of the setts appearing in the source data from MAFF's original surveys had been assigned to the different confidence ratings or the categories detailed in Stage 3 of the draft SOP. For the purposes of this element of the audit, therefore, it was necessary to estimate, wherever possible, to which category each sett would have been assigned.
To be meaningful in the context of the original tessellation, assigning the setts to different categories was done on the basis of MAFF's original survey results and sett classifications, not on the basis of any revisions made as a result of the audit survey results. Nevertheless, some attempt was made to categorise newly discovered setts (again reflecting MAFF's original classification wherever possible). For some setts it was simply not possible reliably to estimate which category would have been chosen. One example highlights the various problems associated with retrospectively assigning these setts to categories, particularly where these have been re-classified on the basis of the audit results: The main sett of Group 6 on Figure 1 was a 14-hole sett recorded as likely to constitute a main sett during the audit surveys. Although it was also recorded as a main sett by the MAFF field surveyor, it was presumably down-graded on review. It was confirmed as a main sett by bait-marking although the group was at the edge of the bait-marking 'cluster' and hence less information was gathered with regard to its territory boundaries than for the adjacent groups. A reasonably obvious territory boundary was revealed to the west of the sett by the bait-marking investigation, but this was not as obvious during the earlier audit surveys and apparently less obvious still during MAFF's original survey. How MAFF would have/should have classified this sett on the basis of the SOP is unclear.
Notwithstanding the difficulties described above, an attempt has been made to analyse the distribution of setts with regard to the different tiles and group territories.
Table 2. The numbers of setts correctly and incorrectly assigned to different tiles.
| Main | Annexe | Subsidiary | Outlier | |
|---|---|---|---|---|
| Setts on correct tile | 5* 5(1) |
1* 1(1) |
3* 2(1) 1(2) |
22* 16(1) 6(2) |
| Setts on incorrect tile | 2* 1(2) 1(4) |
|||
| Setts part of 2 groups encompassed by same tile | 4* 1(0) 2(1) 1(2) |
1* 1(1) |
5* 2(1) 3(2) |
7* 2(1) 5(2) |
| Main | Annexe | Subsidiary | Outlier | |
|---|---|---|---|---|
| Setts on correct tile | 5* 5(1) |
2* 2(1) |
|
10* 9(1) 1(2) |
| Main | Annexe | Subsidiary | Outlier | |
|---|---|---|---|---|
| Setts on correct tile | 2* 2(1) |
1* 1(1) |
1* 1(1) |
1* 1(1) |
| Setts on incorrect tile | 1* 1(3) |
3* 3(2) |
4* 1(2) 2(3) 1(4) |
|
| Setts part of 2 groups encompassed by same tile | 4* 2(0) 2(1) |
3* 3(1) |
2* 2(2) |
7* 3(1) 4(2) |
| Main | Annexe | Subsidiary | Outlier | |
|---|---|---|---|---|
| Setts on correct tile | 4* 4(1) |
2* 2(1) |
1* 1(2) |
11* 8(1) 3(2) |
| Setts on incorrect tile | 1* 1(2) |
1* 1(2) |
1* 1(2) |
|
| Setts part of 2 groups encompassed by same tile |
| Main | Annexe | Subsidiary | Outlier | |
|---|---|---|---|---|
| Setts on correct tile | 2* 2(1) |
1* 1(1) |
||
| Setts on incorrect tile | 2* 2(2) |
1* 1(2) |
1* 1(2) |
1* 2(0) 5(2) |
| Setts part of 2 groups encompassed by same tile | 2* 1(0) 1(1) |
1* 1(0) |
1* 1(1) |
|
Notes for Table 2
Asterisked figures in bold = total number of setts
Smaller figures below = number of setts in each assumed category. Category codes shown in parentheses after each sub-total; (0) indicates that a category could not reliably be assigned.
For Figure 4 - :
(i) Since the bait-marking results were slightly equivocal and indicated
that Groups 1 and 2 were 'linked' to some degree, the setts encompassed
by these groups do not appear in the table; similarly a small number of
outlying setts lying in the 'dead space' between Groups 3 and 4 also do
not appear.
(ii) A further subsidiary sett is also not included in the table; originally this would probably have been included in Category 1 but was found to be located very close to a bait-marking-inferred boundary.
(iii) There will almost certainly be setts to the north and south of Group 6 which were not included in the survey and bait-marking 'cluster' which would have been mis-classified. Similarly some undisclosed setts between Groups 1 and 6 would probably be encompassed by the incorrect tile.
For all main setts included in the analysis (32) approximately 55% (18 setts) were encompassed by the 'correct' tile. Approximately 30% (10 setts) were one of a pair of social groups more or less encompassed by a single tile and only a little over 10% (4 setts) fell within an 'incorrect' tile. As expected, all main setts assigned to the 'correct' tile would have also been assigned to Category 1. Of those main setts not within the 'correct' tile, only 4 would have been included in Category 1 and all of those were one of a pair encompassed by a single tile. All main setts which fell into 'incorrect' tiles would have originally been assigned to Categories 2 or 3.
For all other setts included in the analysis (103) a surprisingly similar proportion (approximately 55%; 56 setts) fell within the 'correct' tile. Approximately 25% (27 setts) formed part of one of a pair of groups covered by the same tile, and approximately 20% (20 setts) fell within an 'incorrect' tile . Approximately 80% (45 setts) included within the 'correct' tiles would also have been assigned to Category 1. In contrast (as with the main setts), all of the setts falling into 'incorrect' tiles would have been assigned to Categories 2, 3 or 4 (or their categories could not reliably be determined).
As detailed above, in the case of the cluster of Groups shown on Figure 5, there were significant discrepancies in main sett classification between the re-survey and MAFF's original survey, and hence significant variations between the tessellation- and bait-marking-derived territories. Figure 6 shows the results of a revised tessellation based on the main setts identified during the re-survey. As with Figure 2, this reveals a general similarity between the newly-tessellated territory boundaries and those derived from the bait-marking results: a total of approximately 65% of the bait-marking boundaries roughly coincide with the tessellated lines; 70% of the areas of each territory (for which an area comparison can be made) coincide; and 100% of the setts fall within the 'correct' territories. It is likely that this agreement would be improved significantly by the correct identification of main setts to the north-west and south-east (outside this particular bait-marking cluster).
It is acknowledged that the tessellations shown on these figures represent some of the earlier attempts, that they have only undergone minor review/revision, and that more recent attempts might be expected to fit 'real' territories more closely. However, it is likely that the variations in main sett classification would undermine most attempts at revision and represent a fundamental stumbling-block with regard to the accuracy of the tessellation boundaries. Notwithstanding these discrepancies, the more detailed analysis of how setts might be assigned to the different tiles has shown that the existing approach 'works' for at least half of the setts, and a more significant proportion if it is accepted that occasionally two nearby groups will be subsumed within the same tile. Also the assignment of setts to the different categories does tend to highlight correctly those setts for which assignment to a given tile is a less reliable indicator of to which group/territory it actually belongs.
3.3 Recommendations
Given how sensitive the social group delineation procedure is to accurate main sett classification, all of the elements of Section 2.3 which relate to improvements in the reliability of sett classification would be expected to have a significant effect on the accuracy and reliability of territory delineation and the correct assignment of setts to groups/territories.
3.3.1 Application of the SOP
The majority of the SOP itself appears to be sound, and the results presented for the cluster shown in Figure 2 and the example re-tessellation shown in Figure 6 indicate that the technique can generate territories similar (and in some cases very similar) to those defined on the basis of the best bait-marking data available.
However, one key stage in the SOP (Items 9 and 10), where the tessellations are intended to provide a starting point for evaluating social group structures within the trial area (and an insight into: the reliability of sett classification; the possible location of setts in areas not surveyed; key areas of badger activity not surveyed; and voids in badger spatial structure) simply did not appear to have been a significant part of the procedure for the areas covered by this audit. (Also, in light of the re-survey results, the tessellations would be extremely helpful in identifying areas within which setts may have been missed).
As recommended in Item 10 of the SOP, some repeat surveys/sett inspections (often by supervisory staff) had taken place in order to clarify sett classification, and possibly to ascertain whether setts were present in areas not surveyed and to confirm the absence of setts in areas which appeared to be voids (it was unclear how much, if any repeat survey work was carried out to confirm the latter). However, it was not clear that these repeat surveys had been instigated by a consideration of the tessellation results, but rather by a routine review of the field survey maps. An analysis of the final tessellation boundaries for each area revealed a significant number drawn as shapes which clearly would not be expected to conform to natural social group territory parameters (one tessellation-derived territory on Figure 4 extended for in excess of 4 km along its long axis), and a significant number encompassing areas far in excess of that which would be expected for a social group territory in the habitat in question. Section 2.3 reinforces how important this continual re-examination of the tessellated boundaries is as part of an iterative approach to the surveying of each triplet area.
3.3.2 Topographic features
With regard to the adjustment of tessellated boundaries (Item 11 of the SOP), it is important that if topographic features are to be used as inferred boundaries, they really should be 'impassable'. In addition, the guidelines in this section suggest that "where main setts are close together (200m or less), and there is no evidence of a border between them, territories should be amalgamated....". It is recommended that this, as with other apparent anomalies revealed by the tessellation procedure, should trigger a re-survey of the area in question (at the very least a re-inspection of the setts involved), and the tessellation amended as indicated by the results of this work.
3.3.3 Training and further audit
It is recommended that further training, and a more formal approach to how and when the tessellation and review procedures are applied, as part of an iterative approach to survey and assessment of the Treatment Areas, be developed. As identified at the end of Section 2.3 (above), it would also be helpful to repeat some element of the Audit to date in a new Treatment Area, to evaluate the improvements to the application of this SOP.
4.0 Acknowledgements
I would like to take this opportunity to record my gratitude to MAFF staff all of whom have been extremely helpful and forthcoming throughout this audit.
Page last modified:
12 August 2003
Page published: 5 February 2003