APPENDIX
DETAILED RESPONSES TO KING ET AL (2007)
(BY PARAGRAPH NUMBER)
PARAGRAPH 5
A1 King et al (2007) refer to the
need for an intervention that would "reduce the prevalence
of disease in both cattle and wildlife". Detailed analysis
of the prevalence of M. bovis infection among badgers culled in
the RBCT demonstrated that prevalence was increased following
both proactive and reactive culls, where prevalence is the proportion
affected (Bourne et al, 2007; Woodroffe et al, 2006b).
We note that Sir David King's oral evidence to the Environment,
Food and Rural Affairs Committee suggests that his team were envisaging
a policy entailing reductions in badger density (ie the numbers
of badgers per unit area) similar to those achieved in the RBCT
(Q407, Environment, Food and Rural Affairs Committee, 2007). In
this context, and in the absence of any data on the level to which
population densities would need to be suppressed in order to reduce
disease transmission among badgers in TB-affected areas of Britain,
we do not consider badger culling able to fulfil King et al
(2007) criteria for appropriate intervention.
A2 Proactive culling, as conducted in the
RBCT, substantially reduced the density of badgers (Woodroffe
et al, 2007), and thus it probably reduced the density
of M. bovis infected badgers despite increased prevalence. However
localised reactive culling produced smaller reductions in overall
badger density (Woodroffe et al, 2007) yet was still associated
with elevated prevalence (Bourne et al, 2007); hence the
density of infected badgers may not have been reduced and could
have been increased.
CONCLUSIONS—THIRD
BULLET POINT
A3 We believe that clarity is critical to
this scientific debate, and therefore note that the statement
that "removal of badgers is the best option... to reduce
the reservoir of infection in wildlife" is imprecise and
potentially misleading. "Reduc[ing] the reservoir of infection
in wildlife" could refer to reducing wildlife density, to
reducing the prevalence of infection within the wildlife population,
or to both. If this wording is intended to refer to a reduction
in the prevalence of infection in wildlife, then it is erroneous:
as described in paragraphs ISG25 and A1 above, there is strong
evidence showing that RBCT culling increased, rather than reduced,
the prevalence of infection in badgers (Bourne et al, 2007;
Woodroffe et al, 2006b). If the wording is intended to
refer to a reduction in the density of badgers, it is important
to recognise the very strong evidence that such a reduction could
decrease or increase TB risks to cattle (or both, Donnelly et
al, 2006; Donnelly et al, 2003), depending on the form
of culling. There is consistent evidence that if badger culling
were to be conducted in a patchy, inefficient, or uncoordinated
manner, then subsequent risks of transmission to cattle would
probably be no lower—and could possibly be higher—than
they were before culling even though badger densities were being
suppressed (Bourne et al, 2007).
CONCLUSIONS—FIFTH
BULLET POINT
A4 We are unsure of the evidence used by
King et al (2007) to select 100 km2 as the minimal area
for badger removal. While this was the scale of RBCT areas, it
was clear from our analyses that, when surrounding areas were
also considered, the benefits within the 100 km2 culled area were
largely offset by the detrimental effects on surrounding land
(Donnelly et al, 2007). Two extrapolations that we considered
could be supported by the RBCT data indicated that the minimal
area required to obtain a statistically significant overall benefit
was either 265 km2 (Paragraph 5.41 and Figure 5.4A in Bourne et
al, 2007) or 455 km2 (Paragraph 5.42 and Figure 5.4B in Bourne
et al, 2007), depending on the assumptions used. Any such
extrapolation requires untestable assumptions (by the very definition
of extrapolation), yet it is clear from the RBCT data that 100
km2 is too small an area to be confident of an overall beneficial
effect on cattle TB incidence.
CONCLUSIONS—NINTH
BULLET POINT
A5 As noted in paragraphs ISG13-ISG15 above,
we strongly disagree with King et al(2007) statement that
the evidence for a detrimental effect is limited to the area between
0.5 and 1 km outside the removal area.
A6 Further, on the basis of RBCT and other
data we consider it highly unlikely that the "soft boundaries"
mentioned would either "reduc[e] the migration of badgers
into the removal area" as stated, or markedly reduce detrimental
effects on neighbouring land. These arguments are outlined in
paragraphs ISG19 and A34-A35 of this document.
CONCLUSIONS—TENTH
BULLET POINT
A7 While King et al (2007) recommend
monitoring of cattle TB incidence in the proposed culling areas,
we are concerned that no detailed consideration has been given
to the precision of the analysis of such monitoring data. Without
knowing the spatial extent over which any such culling would occur,
it is impossible to provide statistical guidance. More importantly,
it would be extremely difficult to obtain reliable estimates of
the effects of culling on TB incidence (particularly any detrimental
effects in neighbouring areas) if no unculled comparison areas
were available. Indeed, this is one of the reasons why Krebs et
al (1997) recommended that a randomised controlled experiment
be conducted, on an appropriate temporal and geographical scale,
to evaluate the impacts of badger culling on cattle TB.
PARAGRAPH 11
A8 As noted in paragraphs ISG20-ISG22 above,
the results of Cox et al (2005) have been misinterpreted.
PARAGRAPH 14
A9 King et al (2007) comment that
"...cattle movements alone cannot explain the persistence
of the geographically compartmentalised areas of high incidence
of cattle TB, nor their gradual expansion over the last decade..."
but provide no evidence in support of this statement. In considering
the potential contribution of cattle movements to the geographical
spread of TB, it is worth noting that, in western England and
Wales, 43% of cattle movements occur over a distance of less than
20 km (Mitchell et al, 2005). Hence, cattle movement is
likely to contribute to local as well as long-distance spread
of infection.
PARAGRAPH 24
A10 This paragraph notes that "most
of the areas had only four removal operations". This is incorrect.
The number of proactive culls conducted in each triplet were:
A five culls, B seven culls, C six culls, D four culls, E six
culls, F five culls, G five culls, H five culls, I four culls,
J four culls, giving a mean of 5.1 and a mode of five culls per
triplet.
A11 Consideration is given in this paragraph
to the non-significant trend for the benefits of culling within
trial areas to increase on later culls. While mathematical modelling
could be used to extrapolate the likely benefits of conducting
further annual culling operations, this would again require untestable
or uncertain assumptions (including, but not limited to, how badger
recolonisation and birth rates might change with repeated culling,
what proportion of breakdowns were due to badger-to-cattle transmission,
and how badger ranging behaviour might change with badger density).
While detailed mathematical models of the cattle-badger disease
system have been constructed (eg Smith et al, 2001), considerable
detailed sensitivity analyses of parameter values and model structure
would be required before their predictions of future benefits
could be considered reliable.
PARAGRAPH 25
A12 The RBCT did demonstrate that repeated
culling carried out by skilled members of the Defra Wildlife Unit
delivered significantly decreased TB incidence in cattle within
the 100km2 RBCT trial areas (Donnelly et al, 2006). However,
these beneficial effects were accompanied by detrimental effects
in neighbouring areas (Donnelly et al, 2006).
A13 With regard to extrapolation of effects
of culling over larger areas, paragraphs 5.41-5.42 and Figure
5.4 of Bourne et al (2007) provide what we consider to
be reliable extrapolations from our results.
PARAGRAPH 27
A14 We are unsure why the increased incidence
of cattle TB observed on land neighbouring RBCT trial areas was
judged "hard to interpret". We have published a body
of research on this issue and the evidence is considered "substantial"
and "strong" by scientific authorities (eg Royal Society,
2006; Shepherd, 2005; Shepherd et al, 2005). We have published
a number of peer-reviewed papers showing that culling leads to
disrupted territorial organisation and expanded ranging by badgers
(Pope et al, 2007b; Woodroffe et al, 2006a), as
well as immigration of badgers into culled areas (Woodroffe et
al, 2007). These ecological changes are associated with reduced
clustering of infection in both badgers and cattle (Jenkins et
al, 2007), and with increased prevalence of infection in badgers
(Bourne et al, 2007; Woodroffe et al, 2006b). The
causality of the relationship between culling-induced changes
in badger ecology and increased TB incidence in cattle is suggested
not only by these broad patterns, but also by details including
modification of the effects by factors such as geographic barriers
to badger movement, and the practice of non-simultaneous culling
(Woodroffe et al, 2006b). Indeed, it was the observation
of expanded ranging by badgers living immediately outside proactive
culling areas (Woodroffe et al, 2006a) that prompted us
to investigate cattle TB incidence on these lands (Donnelly et
al, 2006). It is unfortunate that none of this ecological
work is cited by King et al (2007).
A15 King et al (2007) go on to recommend
that "measures... should be put in place to minimise that
increase [in cattle TB]". This appears to ignore the detailed
consideration in paragraphs 10.8-10.48 of Bourne et al
(2007) which concludes that no practical measures are likely to
be able to achieve this. Further details are given in paragraphs
A33-A35 below.
PARAGRAPH 28
A16 As mentioned in paragraph ISG16 above,
we consider it inappropriate to exclude data from the period between
the first and second culls. The reason for this is detailed in
the peer-reviewed Supplementary Information of Donnelly et
al (2006), which we reproduce here.
INTERPRETATION OF
ANALYSES FROM
DIFFERENT TIME
PERIODS
The main text presents analyses from two time
periods, one dating from the completion of the initial cull (which
shows statistically significant effects both inside and outside
trial areas), and one dating from completion of the first follow-up
cull (which shows a significant effect inside, but a non-significant
trend outside). The reasons for considering these two time periods,
and their implications for the interpretation of our findings,
merit further comment.
We used incidence from the date of the initial
cull as our primary analysis, mainly because this measure is the
most relevant to policy: the effects detected reflect what one
could expect to achieve from a proactive culling policy implemented
on the timescale measured. As described briefly in the main text,
there were two reasons for performing secondary analyses which
excluded data from before the first follow-up cull. First, this
excluded breakdowns that might have originated prior to the onset
of culling, even though (given annual testing) they were not detected
until after culling had begun. Such breakdowns would lead to under-estimation
of culling-induced effects on incidence in the first year—essentially
this would bias estimates of increases or reductions toward no
effect. Our findings of statistically significant effects dating
from completion of the initial cull, both inside and outside trial
areas, therefore indicate the strength of both positive and negative
effects of culling.
An additional reason for performing the secondary
analyses was that a more complete badger removal would have been
achieved from the date of the first follow-up cull. This more
complete cull would be expected to generate a greater reduction
in cattle TB inside trial areas, and the results of the secondary
analysis are indeed consistent with this prediction, albeit with
a wider confidence interval due to the smaller dataset.
In contrast with the situation inside trial
areas, however, the circumstances of incomplete badger removal
that would have occurred between the initial cull and the first
follow-up could be expected to increase any detrimental effects
of culling, if such effects were caused by disruption of badger
territorial organisation at artificially reduced population densities.
The frequency of potentially infectious contacts between cattle
and badgers will be related to both the density of badgers, and
the ranging behaviour of those badgers. We have hypothesised that,
where densities are substantially reduced, contact rates will
be reduced despite expanded ranging behaviour, but that smaller
reductions in density will generate increased contact rates if
(as observed) they are also accompanied by expanded ranging. In
this scenario, we would expect detrimental effects to be particularly
marked following the initial cull since densities were probably
reduced to a lesser extent during this period. Our secondary analysis
excluded this potentially important time period and this, along
with the reduced sample size, helps to explain why the culling
effect in `neighbouring areas' was found to be weaker.
PARAGRAPH 29
A17 As noted by King et al (2007),
"the detrimental effect was not spread over all of the area
[up to] 2km outside the removal area" although we wish to
clarify that the analyses to which this comment refers considered
distance from the boundary of the trial area, rather than the
treatment area (Donnelly et al, 2007). Trial area boundaries
were delineated mainly along property boundaries, so that herds
could in principle be classified unambiguously as located inside
or outside the trial area. Treatment areas, within which culling
was conducted, were slightly larger than trial areas, and delineated
according to the estimated boundaries of social group territories
so that all badgers using farms inside the trial areas could be
targeted (Bourne et al, 2007). As a consequence, some culling
was conducted on land immediately adjoining trial areas, but outside
their boundaries, and this almost certainly explains the (non-significant)
beneficial effect of proactive culling among herds on land 0-0.5
km outside RBCT trial areas.
A18 Our decision to analyse data from all
land up to 2 km outside trial areas (except when it was within
2 km of more than one trial area), rather than treatment areas,
was a deliberate one, taken to be conservative and to avoid any
accusations that data were selected or excluded in order to obtain
a particular result. However, the occurrence of apparently beneficial
effects within 500m of the trial area boundary would be expected
to offset, to some extent, the detrimental effects observed at
greater distances and could therefore lead to under-estimation
of detrimental effects on unculled land.
A19 To provide further clarity on this point,
we have repeated our analyses of the incidence of cattle TB on
land up to 2km outside trial area boundaries, excluding herds
occupying land outside the trial area but inside the treatment
area. These analyses, which are presented in Table 1 below, reveal
estimated detrimental effects slightly stronger than those reported
in our published papers. These results confirm that our inclusion,
in primary analyses, of herds outside the trial areas but inside
the treatment areas was conservative and probably led to slightly
under-estimation of detrimental effects on unculled land.
A20 As discussed in paragraphs ISG13-ISG15
above, it is inappropriate to judge the overall detrimental effect
of culling by considering confidence intervals for individual
subgroups of data. Also, as noted in paragraphs ISG16 and A16
above, and in paragraphs A40-A44 below, we consider it appropriate
to include all data, from the completion of the initial proactive
cull, in primary analyses of the effects of culling.
Table 1
Comparison of estimated detrimental effects
of proactive badger culling on the incidence of cattle TB on farms
up to 2km outside RBCT areas, when herds falling outside the trial
areas but inside the treatment areas are included (as in our published
analyses, Bourne et al, 2007; Donnelly et al, 2007)
or excluded. Confidence intervals and p values are adjusted for
overdispersion as described in our published work.
|
| Including herds inside treatment area
| Excluding herds inside treatment area
|
| Time period | estimated
effect
| (95% CI) | p
| estimated
effect
| (95% CI) | p
|
|
| Using VetNet location data |
| | | |
| |
| First cull to one year after last cull | 24.5%
| (-0.6-56.0%) | 0.057
| 29.3% | (1.1-65.3%)
| 0.040 |
| Second cull to one year after last cull |
19.6% | (-10.3%-59.5%)
| 0.22 | 23.1%
| (-7.9-64.6%) | 0.16
|
| First cull to second cull | 46.8%
| (-0.4-116.4%) | 0.052
| 57.0% | (-4.3-157.5%)
| 0.074 |
| Using RBCT location data | |
| | |
| |
| First cull to one year after last cull | 35.3%
| (5.8-73.0%) | 0.016
| 43.8% | (9.1-89.6%)
| 0.010 |
| Second cull to one year after last cull |
24.9% | (-7.2-67.9%)
| 0.14 | 35.5%
| (-1.1-85.5%) | 0.058
|
| First cull to second cull | 95.4%
| (10.5-245.5%) | 0.021
| 95.0% | (3.3-264.3%)
| 0.039 |
| | |
| 57.0% | (-4.3-157.5%)
| 0.074 |
|
PARAGRAPH 30
A21 King et al (2007) found it "hard to compare
data" from our report as "figures are presented as percentages
rather than as absolute numbers of herd breakdowns". We are
surprised at this concern. It is standard to report percentage
differences in risks, and then for these to be translated into
absolute numbers for specific considerations, such as cost benefit
analyses. Indeed, we give specific calculations for 100 km2 areas
in Bourne et al (2007) paragraph 5.39. These indicate that,
over a five-year period of culling, 116 fewer confirmed breakdowns
would have occurred within 10 circular 100 km2 areas, and 102
additional breakdowns would have occurred in the 10 associated
83.5 km2 neighbouring areas. This gives a net overall benefit
of 14 fewer confirmed breakdowns (Donnelly et al, 2007).
Detailed consideration of the costs and benefits of this scale
of operation was presented in Chapter 9 of Bourne et al
(2007).
A22 King et al (2007) suggest that the potential
role of cattle herd densities was not considered in our analyses.
In fact, the inclusion of the (ln transformed) number of herds
as an independent variable in our log-linear regression analyses
allowed us to characterise the relationship between herd density
and the incidence of cattle TB, since all trial areas were of
similar size.
A23 The full raw data on the numbers of herds, breakdowns
and historic breakdowns were provided in supplementary data files
published with both papers on the effects of proactive culling
(Donnelly et al, 2007; Donnelly et al, 2006). While
we did not specifically report herd densities (per km2, for example)
in Bourne et al (2007), we did provide the number of cattle
herds at baseline in Tables 5.1 and 5.7, and the sizes of trial
areas are given in the peer-reviewed Supplementary Information
of Donnelly et al (2006). Hence, relevant data have been
publicly available since December 2005 (when Donnelly et al
(2006) was published online), allowing interested parties to examine
herd density effects in further detail had they so wished.
PARAGRAPH 31
A24 As noted in paragraphs A33-A35 below, King et
al (2007) comment that "the overall beneficial effect
on incidence of cattle TB will be maximised if steps are taken
to minimise that detrimental effect" fails to take account
of the absence of practical measures likely to minimise detrimental
effects.
A25 As noted in paragraph A7 above, we are concerned
that it would be extremely difficult to obtain reliable estimates
of the effects of badger culling on the incidence of cattle TB
by "monitoring of these effects up to 2 km outside the removal
area" if no comparable unculled comparison areas were available.
Indeed, this is one of the reasons why Krebs et al (1997)
recommended that a randomised controlled experiment be conducted
to measure the impact of badger culling on cattle TB.
PARAGRAPH 33
A26 We consider the evidence for the reduction in badger
density achieved through proactive culling to be considerably
stronger than "an informed guess". Multiple indices
of badger abundance gave similar estimates of the impact of proactive
culling (Woodroffe et al, 2007). Several of these indices
were based on densities of badger field signs; these were shown
to correlate with the numbers of badgers captured per unit area
on initial culls, indicating that they are likely to be reasonable
measures of badger abundance (Woodroffe et al, 2007). Another
index used, the density of road-killed badgers retrieved, is a
more direct measure of badger abundance and gave similar results
(Woodroffe et al, 2007). Improved methods for measuring
badger density were developed while the RBCT was in progress (Frantz
et al, 2004; Hounsome et al, 2005; Wilson et
al, 2003) but were not available at the start of the study
and so could not be used to monitor culling effects throughout
the course of the study. Nevertheless we consider the close agreement
between estimates based on different measures to indicate that
the effects of culling on badger density were estimated reasonably
reliably (Woodroffe et al, 2007).
PARAGRAPH 36
A27 King et al (2007) rightly note that clustering
of M. bovis infection in badgers "was disrupted over the
course of the trial and... the prevalence of infection in badgers...
increased". Their report fails to note, however, that similar
changes in the clustering of infection were also observed in cattle:
in reactively culled areas, and on land neighbouring proactively
culled areas, cattle infections became less clustered as successive
badger culls were conducted (Jenkins et al, 2007). This
change in the geographical distribution of cattle infections reflects
expanded ranging behaviour by badgers in the same areas (Woodroffe
et al, 2006a), contributing to the large body of evidence
linking detrimental effects of culling to disruption of badger
social organisation.
PARAGRAPH 37
A28 The conclusions reached in this paragraph are difficult
to interpret since it is not clear whether they refer to badger
populations inside the culling areas, to those immediately outside,
or to both. In either case, the statement that "The ISG considered
that the disruption of badgers and the increased ranging behaviour
was a permanent effect... However, there is a reasonable possibility
that the disruption is transient" is misleading in that it
implies that the ISG gave no consideration to temporal trends.
In fact, Woodroffe et al (2006b), Woodroffe et al
(2007) and Pope et al (2007a) sought evidence for temporal
trends in the effects on badgers, Donnelly et al (2007)
and Bourne et al (2007) sought evidence for temporal trends
in the beneficial and detrimental effects on cattle, and Jenkins
et al (2007) evaluated temporal trends in the relationship
between infections in the two host species. In particular, the
Discussion section of Donnelly et al (2007) includes several
paragraphs proposing an ecological mechanism that might explain
the temporal trends observed in cattle.
A29 The first investigation of any transience of the
effects on cattle was not a "simple regression" as stated,
but rather a detailed consideration of the stratified data presented
in Figure 5.2A of Bourne et al (2007). Any modelling beyond
the linear regression suggested by the pattern observed over the
first four culls would require untestable assumptions and would
thus be unlikely to give consistent results over the possible
range of parameter values and model structures consistent with
the known data. The evidence of the RBCT is clear over the first
four annual culls.
A30 While King et al (2007) are correct that "the
data do not discount this theory" [that disruption of badger
spatial organisation is transient] outside culling areas where
data are limited, data from inside culling areas suggest that
disruption was sustained. As successive proactive culls were conducted,
an increasing proportion of badgers were captured close to culling
area boundaries, indicating sustained badger immigration into
culled areas with no evidence of this effect levelling off on
later culls (Woodroffe et al, 2007). Genetic studies likewise
show an increasing proportion of the badger population engaging
in long-distance movements following successive culls, confirming
sustained disruption of badger populations (Pope et al,
2007a). Finally, the prevalence of M. bovis infection among badgers
rose, and the spatial distribution of infection became less clustered,
on successive culls, once again with no evidence of lesser effects
on later culls (Jenkins et al, 2007; Woodroffe et al,
2006b) despite declining badger density over the same time period
(Woodroffe et al, 2007). All of this evidence indicates
that "disruption of badgers" was sustained throughout
the course of the RBCT, contrary to King et al's suggestion.
In contrast, evidence cannot discount the hypothesis that detrimental
effects for cattle may have declined on later proactive culls
(Donnelly et al, 2007) as the suppression of badger densities
was sustained. The reduction in spatial clustering of cattle infections
likewise appears to have been greatest between the first and second
culls (Jenkins et al, 2007).
PARAGRAPH 38
A31 As in paragraph 37 of King et al (2007), it
is not clear whether the conclusions reached in this paragraph
refer to badger populations inside, or immediately outside, the
culling areas. If they refer to badgers inside the culling area
then, as detailed above, the statement that "the likelihood
of uninfected badgers being exposed to infectious badgers will...
be reduced, ideally to a level at which TB cannot sustain itself
within the badger population" is at odds with the available
evidence, which has been published in peer-reviewed scientific
journals. Woodroffe et al (2006b) showed that the prevalence
of infection among badgers increased as successive proactive culls
depressed badger density (Woodroffe et al, 2007). Moreover,
the finding that these infections were also less spatially clustered
on later culls (Jenkins et al, 2007) is consistent with
the hypothesis that disruption of badger spatial organisation
led to increased badger-to-badger transmission (Woodroffe et
al, 2006a). While in principle it is to be expected that extremely
low population densities would limit contact between badgers and,
hence, badger-to-badger transmission of infection, there are no
data to indicate the level of population reduction needed to achieve
this in TB-affected areas of Britain. Moreover, there is evidence
for widespread cattle-to-badger transmission in Britain (Jenkins
et al, 2007; Woodroffe et al, 2006b), and this could
well contribute to continued infection in very low-density badger
populations despite low badger-to-badger transmission rates.
A32 As noted in paragraphs A2 and A3 above, substantially
lowering badger density by sustained, simultaneous, coordinated
culling across very large areas could reduce the incidence of
cattle TB inside culled areas, even though the remaining badgers
might experience increased prevalence of infection. However, since
the relationship between badger density and TB risk to cattle
is strongly non-linear, culling in smaller areas, for shorter
time periods, or in an uncoordinated manner will all seriously
undermine any beneficial effects of culling and have the potential
to generate detrimental effects.
PARAGRAPH 39
A33 All of the measures proposed in this paragraph were
considered systematically in Chapter 10 of Bourne et al
(2007), and found to be unworkable. For example, we noted that
too few existing barriers to badger movement occur in southern
and western Britain for these to be used to delineate culling
zones over a meaningful proportion of TB-affected areas, and that
the mitigating effect of such barriers on the incidence of cattle
TB (as opposed to infection prevalence in badgers) was unproven.
We also drew attention to the cost of constructing badger-proof
electric fences (Poole et al, 2002), and the impracticality
of fencing any but the smallest culling areas given the number
of roads traversing the British countryside.
A34 We wish to comment in particular on the suggestion
of using "...soft boundaries such as arable land with no
cattle..." to minimise detrimental effects on cattle TB.
This sort of approach—which we considered in paragraph 10.22
of Bourne et al (2007)—would not prevent disruption
of badger populations immediately outside culling areas (since
badgers regularly occupy arable land). However, this measure could
reduce the impact of such disruption on cattle by ensuring that
few or no cattle inhabited the areas where disease transmission
from badgers was most likely.
A35 In our view, any such "soft boundaries"
would have to be substantially wider than the 1 km proposed. We
observed changes in badger ranging behaviour, and detrimental
effects on cattle TB, up to 2 km outside RBCT trial area boundaries
(Donnelly et al, 2006; Woodroffe et al, 2006a).
Bait-marking studies can be used to derive conservative estimates
of badger home range sizes inside proactively culled areas: the
mean value was 0.77 km2, equivalent to a circle with a diameter
of approximately 1km. Adding one standard deviation to this mean
home range size gives an area of 1.34 km2, equivalent to a diameter
of about 1.3 km. A 1 km-wide band would therefore be fairly small
in relation to the scale of badger ranging, and would be regularly
traversed by local badgers. We therefore considered somewhat wider
bands in Bourne et al (2007). To illustrate the scale of
effects, if culling were to be conducted within circular 100 km2
areas, each surrounded by a cattle-free buffer 2 km wide, each
buffer would cover 83.5 km2. Buffers would be proportionally smaller
(though absolutely larger) for larger culling areas, and would
also be larger where culling areas were not circular. We concluded
that the costs (in the broadest sense) of excluding cattle from
such large areas of the British countryside would be likely to
out-weigh the benefits (Bourne et al, 2007).
PARAGRAPH 42
A36 This paragraph contains a statement fundamental to
King et al's (2007) conclusions, namely that our own "...view
that this benefit [of proactive culling] was largely offset by
the increase in incidence outside the removal area is unsound".
We consider this statement to be inconsistent with the data available.
As discussed elsewhere in this document, King et al's downplaying
of the detrimental effects of culling appears to be based upon
a number of misunderstandings including inappropriate interpretation
of confidence limits (paragraphs ISG13-ISG15), exclusion of data
from the initial time period which cannot be justified on the
basis of any statistical bias toward overestimation (paragraphs
ISG16 and A16) or time delay between performing badger culling
and detecting its effects (paragraphs A40-A44), and failure to
take full account of ecological data which offer a plausible and
consistent explanation for both the detrimental and beneficial
effects observed (paragraph ISG18). In particular, the increased
incidence of cattle TB observed in herds up to 2 km outside proactively
culled trial areas was a consistently observed phenomenon and,
given its magnitude, largely offset the benefits of reduced TB
incidence among cattle within proactively culled trial areas (Donnelly
et al, 2007). This offsetting is clearly demonstrated for
culling areas up to 300 km2 in Figure 5.4 of Bourne et al
(2007).
PARAGRAPH 43
A37 As mentioned in paragraphs ISG13-ISG15 above, counting
how many confidence intervals of stratified analyses include zero
is statistically inappropriate.
A38 The occurrence of some badger removal immediately
outside RBCT trial areas, and the way in which data from these
areas were included in analyses, should not have been "unclear"
given the detailed descriptions of methodology provided in Bourne
et al (2007). The "trial" and "treatment"
areas are defined in paragraphs 2.11 and 2.12, and illustrated
in Figure 2.1. Paragraphs 5.5 and 5.25 indicate that primary analyses
concerned herds inside or outside trial (rather than treatment)
areas and paragraphs 5.15 and 5.33 make explicit reference to
distances from the "trial area boundary". Moreover,
paragraph 5.33 states that "...herds... less than 0.5 km
outside the trial area boundary appeared to experience a benefit...
this... was unsurprising, because badger culling extended just
beyond the boundaries of the trial areas to target social groups
judged... to occupy home ranges falling partially inside the trial
areas...".
A39 To provide further clarity on this point, we have
repeated our analyses of the incidence of cattle TB on land up
to 2km outside trial area boundaries, excluding herds occupying
land outside the trial area but inside the treatment area. These
results are presented in paragraph A19 above.
PARAGRAPH 44-47
A40 King et al (2007) note that "it would
be reasonable to expect... a time lag between removal of badgers
and detection of changes in infection in cattle" but that
"this time lag does not seem to have been taken into account
when the ISG collected data on cattle TB incidence immediately
after the first proactive removal".
A41 Our reasons for including data between the first
and second proactive culls in our primary analyses are detailed
in paragraphs ISG16 and A16 above. Since we considered it likely
that the effects of culling might change over time, our publications
also presented estimates of effects from the date of completion
of the second cull (Bourne et al, 2007; Donnelly et
al, 2007; Donnelly et al, 2006) and explicitly investigated
changes across different time periods (Bourne et al, 2007;
Donnelly et al, 2007). The suggestion that a possible time
lag "does not seem to have been taken into account"
is therefore incorrect.
A42 We discussed this possible time lag in detail in
the peer-reviewed Supplementary Information of Donnelly et
al (2006). Available data indicate that such time lags could
be short. This is because (i) behavioural data show that local
reductions in badger density affect ranging behaviour within a
few days or weeks (Cheeseman et al, 1993; Roper & Lups,
1993; Woodroffe, Macdonald & da Silva, 1995), allowing contact
with additional cattle herds; and (ii) once infected, cattle become
responsive to the tuberculin test after approximately three weeks
(Thom et al, 2006). Hence, if badgers can infect susceptible
cattle rapidly on contact, increased cattle incidence would be
detectable two to three months after badger culling.
A43 King et al (2007) state that naturally-acquired
infections entail longer delays to skin test responsiveness than
do experimental infections, but provide no data to support this
assertion. We are not persuaded that time to responsiveness could
be estimated for natural cases since infection dates would be
unknown. Thom et al (2006) used infective doses of M. bovis
that resulted in disease similar to that observed in naturally
infected cattle, and we therefore consider their experimental
findings the most reliable data currently available.
A44 We also wish to note that this concern about time
delays refers to our hypothesis about the mechanism whereby badger
culling prompts detrimental effects in cattle, not to the existence
of detrimental effects themselves. Since there is strong and highly
consistent evidence that detrimental effects occur, and since
these are costly for farmers, for the farming industry, and ultimately
for the taxpayer, it is vital that they be taken into account
in developing TB control policy.
PARAGRAPH 48
A45 As discussed in paragraphs ISG18-ISG19 and A14 above,
we consider it unfortunate that King et al (2007) state
that they were "not fully persuaded by" our explanation
for detrimental effects of culling yet fail to cite any of our
peer-reviewed papers which provide strong support for this hypothesis.
As noted in paragraph A14, there is consensus within the scientific
community that evidence in support of our hypothesis is "substantial"
and "strong" (eg Royal Society, 2006; Shepherd, 2005).
PARAGRAPH 49
A46 As detailed in paragraphs A28-A30 above, the ISG
gave explicit consideration to the possibility that the effects
of culling might change over time.
PARAGRAPH 50
A47 King et al's (2007) statement that RBCT data
should not be used "...to either support or rule out a reactive
removal strategy..." contrasts with their earlier conclusion
that "...the minimum overall area within which badger removal
should take place is 100 km2" (the average area targeted
by reactive culling was 8.8 km2).
A48 King et al (2007) dismiss our findings regarding
reactive culling, partly because this part of the RBCT was "stopped
before robust results could be obtained". We note that the
decision to halt reactive culling was taken by Defra ministers;
the ISG had recommended that culling be continued while recognising
that this might be difficult for Defra to justify (Bourne et
al, 2005).
A49 Despite this, the consistency of our original results
(Donnelly et al, 2003) with the findings of subsequent
analyses indicate that our conclusions concerning reactive culling
are indeed "robust". Subsequent analyses provide (i)
evidence that detrimental effects occur on land neighbouring proactive
culling areas (Donnelly et al, 2006), making an overall
detrimental effect predictable where culling areas are small and
hence exceeded in extent by the areas of neighbouring land (Bourne
et al, 2007); (ii) evidence that the detrimental effect
of reactive culling disappeared following cessation of culling,
indicating that the effect was not due to a systematic bias between
trial areas unrelated to reactive culling (Bourne et al,
2007); (iii) evidence that herds located in close proximity to
reactively culled land experienced elevated TB risk, even after
controlling for the effect of contiguous breakdowns (Bourne et
al, 2007); (iv) evidence that repeated reactive culling was
associated with spatial spread of infection in cattle (Jenkins
et al, 2007); and (v) evidence that repeated reactive culling,
like proactive culling, was associated with elevated infection
prevalence in badgers (Bourne et al, 2007). This information
indicates that it is extremely unlikely that a future reactive
culling strategy could contribute to the control of cattle TB,
and would probably exacerbate disease spread. Given this evidence
we consider it remarkable that King et al (2007) failed
to "rule out a reactive removal strategy".
ANNEX 1
A50 In a comment about the use and interpretation of
confidence intervals, King et al (2007) comment that our
inclusion of "decimal points... may give the impression of
more certainty than is the case." While we agree that the
individual values make sense only to fewer digits, our reason
for giving more in this case had a scientific basis: it was to
allow any reader wishing to make some additional calculations
with the limits to do so without appreciable loss of information
from rounding errors.
ACKNOWLEDGEMENTS
We wish to thank former ISG research assistants Helen Jenkins
and Tom Johnston for conducting the additional analyses presented
in paragraph A19.
REFERENCES
Bourne, J, Donnelly, C, Cox, D, Gettinby, G, McInerney, JP, Morrison,
I, and Woodroffe, R (1998) Towards a sustainable policy to control
TB in cattle—A scientific initiative. First Report of the
Independent Scientific Group on Cattle TB, Ministry of Agriculture,
Fisheries and Food, London.
Bourne, J, Donnelly, CA, Cox, DR, Gettinby, G, McInerney, JP,
Morrison, WI, and Woodroffe, R (2005) An epidemiological investigation
into bovine tuberculosis—towards a science-based control
strategy. Fourth Report of the Independent Scientific Group on
Cattle TB, Department for Environment, Food and Rural Affairs,
London.
Bourne, J, Donnelly, CA, Cox, DR, Gettinby, G, McInerney, JP,
Morrison, WI, and Woodroffe, R (2007) Bovine TB: the scientific
evidence Defra www.defra.gov.uk/animalh/tb/isg/pdf/final—report.pdf,
London.
Cheeseman, CL, Mallinson, PJ, Ryan, J, and Wilesmith, JW (1993).
Recolonisation by badgers in Gloucestershire. In The Badger (ed
TJ Hayden), pp 78-93. Royal Irish Academy, Dublin.
Cox, DR, Donnelly, CA, Bourne, FJ, Gettinby, G, McInerney, JP,
Morrison, WI, and Woodroffe, R (2005) Simple model for tuberculosis
in cattle and badgers. Proceedings of the National Academy
of Sciences of the United States of America, 102, 17588-17593.
Donnelly, CA, Wei, G, Johnston, WT, Cox, DR, Woodroffe, R, Bourne,
FJ, Cheeseman, CL, Clifton-Hadley, RS, Gettinby, G, Gilks, P,
Jenkins, HE, Le Fevre, AM, McInerney, JP, and Morrison, WI (2007)
Impacts of widespread badger culling on cattle tuberculosis: concluding
analyses from a large-scale field trial. International Journal
of Infectious Disease, 11, 300-308.
Donnelly, CA, Woodroffe, R, Cox, DR, Bourne, FJ, Cheeseman, CL,
Wei, G, Gettinby, G, Gilks, P, Jenkins, H, Johnston, WT, Le Fevre,
AM, McInerney, JP, and Morrison, WI (2006) Positive and negative
effects of widespread badger culling on cattle tuberculosis. Nature,
439, 843-846.
Donnelly, CA, Woodroffe, R, Cox, DR, Bourne, J, Gettinby, G, Le
Fevre, AM, McInerney, JP, and Morrison, WI (2003) Impact of localized
badger culling on TB incidence in British cattle. Nature,
426, 834-837.
Environment, Food and Rural Affairs Committee (2007) Uncorrected
transcript of oral evidence taken on Wednesday 24 October 2007.
Frantz, AC, Schaul, M, Pope, LC, Fack, F, Schley, L, Muller, CP,
and Roper, TJ (2004) Estimating population size by genotyping
remotely plucked hair: the Eurasian badger. Journal of Applied
Ecology, 41, 985-995.
Hounsome, TD, Young, RP, Davison, J, Yarnell, RW, Trewby, ID,
Garnett, BT, Delahay, RJ, and Wilson, GJ (2005) An evaluation
of distance sampling to estimate badger (Meles meles) abundance.
Journal of Zoology, 266, 81-87.
Jenkins, HE, Woodroffe, R, Donnelly, CA, Cox, DR, Johnston, WT,
Bourne, FJ, Cheeseman, CL, Clifton-Hadley, RS, Gettinby, G, Gilks,
P, Hewinson, RG, McInerney, JP, and Morrison, WI (2007) Effects
of culling on spatial associations of Mycobacterium bovis infections
in badgers and cattle. Journal of Applied Ecology, 44,
897-908.
Keeling, MJ and Eames, KTD (2005) Networks and epidemic models.
Journal of the Royal Society Interface, 2, 295-307.
King, D, Roper, TJ, Young, D, Woolhouse, MEJ, Collins, DA, and
Wood, P (2007) Tuberculosis in cattle and badgers: a report by
the Chief Scientific Adviser, Sir David King http://www.dius.gov.uk/publications/experts%20advice%20on%20badgers%20and%20cattle%20TB.pdf,
London.
Krebs, JR, Anderson, R, Clutton-Brock, T, Morrison, I, Young,
D, Donnelly, C, Frost, S, and Woodroffe, R (1997) Bovine tuberculosis
in cattle and badgers HMSO, London.
Mitchell, A, Bourn, D, Mawdsley, D, Wint, GRW, Clifton-Hadley,
RS, and Gilbert, M (2005) Characteristics of cattle movement in
Britain—an analysis of records from the Cattle Tracing System.
Animal Science, 80, 265-273.
Poole, DW, McKilliop, IG, Western, G, Hancocks, PJ, and Packer,
JJ (2002) Effectiveness of an electric fence to reduce badger
(Meles meles) damage to field crops. Crop Protection, 21,
409-417.
Pope, LC, Butlin, RK, Wilson, GJ, Erven, K, Conyers, CM, Franklin,
T, Delahay, RJ, Cheeseman, CL, and Burke, TA (2007a). A molecular
genetic analysis of badger social structure and bovine tuberculosis—final
report on project SE3110. Defra, London.
Pope, LC, Butlin, RK, Wilson, GJ, Erven, K, Conyers, CM, Franklin,
T, Delahay, RJ, Cheeseman, CL, Woodroffe, R, and Burke, TA (2007b)
Genetic evidence that culling increases badger movement: implication
for the spread of bovine tuberculosis. Molecular Ecology,
doi: 10.1111/j.1365-294X.2007.03553.x.
Roper, TJ and Lups, P (1993) Disruption of territorial behaviour
in badgers Meles meles. Zeitschrift fur Saugetierkunde,
58, 252-255.
Royal Society (2006) Response to Defra's consultation on controlling
the spread of bovine tuberculosis in cattle in high incidence
areas in England: badger culling http://www.royalsoc.ac.uk/displaypagedoc.asp?id=19414.
Shepherd, J (2005) Independent review of research on bovine tuberculosis—SAC-TB
(05)5 Supplementary Report (final) Defra www.defra.gov.uk/science/documents/
papers/2005/SAC-TB(05)5%20Supplementary%20Report.pdf, London.
Shepherd, J, Gaskell, C, Waage, J, and Beddington, J (2005) Independent
review of research on bovine tuberculosis—SAC-TB(05)4 Final
Report Defra www.defra.gov.uk/science/documents/papers/2005/SAC-TB%20(05)%204%20report%20FINAL.pdf,
London.
Smith, GC, Cheeseman, CL, Wilkinson, D, and Clifton-Hadley, RS
(2001) A model of bovine tuberculosis in the badger Meles meles:
the inclusion of cattle and the use of a live test. Journal
of Applied Ecology, 38, 520-535.
Thom, ML, Hope, JC, McAulay, M, Villareal-Ramos, B, Coffey, TJ,
Stephens, S, Vordermeier, HM, and Howard, CJ (2006) The effect
of tuberculin testing on the development of cell-mediated immune
responses during Mycobacterium bovis infection. Veterinary
Immunology and Immunopathology, 114, 25-36.
Wilson, GJ, Frantz, AC, Pope, LC, Roper, TJ, Burke, TA, Cheeseman,
CL, and Delahay, RJ (2003) Estimation of badger abundance using
faecal DNA typing. Journal of Applied Ecology, 40, 658-666.
Woodroffe, R, Donnelly, CA, Cox, DR, Bourne, FJ, Cheeseman, CL,
Delahay, RJ, Gettinby, G, McInerney, JP, and Morrison, WI (2006a)
Effects of culling on badger (Meles meles) spatial organization:
implications for the control of bovine tuberculosis. Journal
of Applied Ecology, 43, 1-10.
Woodroffe, R, Donnelly, CA, Jenkins, HE, Johnston, WT, Cox, DR,
Bourne, FJ, Cheeseman, CL, Delahay, RJ, Clifton-Hadley, RS, Gettinby,
G, Gilks, P, Hewinson, RG, McInerney, JP, and Morrison, WI (2006b)
Culling and cattle controls influence tuberculosis risk for badgers.
Proceedings of the National Academy of Sciences of the United
States of America, 103, 14713-14717.
Woodroffe, R, Gilks, P, Johnston, WT, Le Fevre, AM, Cox, DR, Donnelly,
CA, Bourne, FJ, Cheeseman, CL, Gettinby, G, McInerney, JP, and
Morrison, WI (2007) Effects of culling on badger abundance: implications
for tuberculosis control. Journal of Zoology, doi:10.1111/j.1469-7998.2007.00353.x.
Woodroffe, R, Macdonald, DW, and da Silva, J (1995) Dispersal
and philopatry in the European badger, Meles meles. Journal
of Zoology, 237, 227-239.
|
