Journal of Environmental Management 70 (2004) 323–332
www.elsevier.com/locate/jenvman
Controlling Rhododendron ponticum in the British Isles:
an economic analysis
Katharina Dehnen-Schmutza,*, Charles Perringsa, Mark Williamsonb
a
Environment Department, University of York, Heslington, York YO10 5DD, UK
Department of Biology, University of York, P.O. Box 373, York YO10 5YW, UK
b
Received 7 July 2003; revised 10 December 2003; accepted 10 December 2003
Abstract
What resources should be committed to the control of invasive species? This study is based on a survey of nature conservation and forestry
authorities, wildlife trusts and private landowners which investigated the extent of the ecological and economic impacts of the invasive nonnative plant Rhododendron ponticum in the British Isles. There are data on 52,000 ha of land affected by R. ponticum, more than 30,000 ha of
it in nature reserves. For nearly all nature reserves, displacement of native species and habitat changes were both reported. In 2001,
respondents controlled 1275 ha of R. ponticum at a cost of £670,924. To test the optimality of this, we apply a model of social expenditure.
The external costs of R. ponticum control are estimated from the probability that it will spread to contiguous sites and the damage done on
invaded sites. These are then used to calculate the socially optimal level of expenditure on R. ponticum control, and the funding gap it
identified by comparing the result with current levels of expenditure. The results suggest that a socially optimal level of control effort requires
a significant increase in social funding for R. ponticum control, although the size of the increase varies between landholders.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: Rhododendron ponticum; Invasive species; Control costs; External costs
1. Introduction
Increasing attention has been paid recently to the
economic consequences of invasive species (US OTA,
1993; Perrings et al., 2000; Pimentel et al., 2000, 2002). The
economic consequences of invasive species include damage
costs, such as biodiversity loss or habitat change, plus the
costs of control or eradication net of any possible benefits.
Whereas damage costs are not directly calculable, the costs
of control and eradication are. The costs of control are a
measure of the effort committed to the eradication or
reduction of an invasive species. Since individual landowners will increase control effort up to the point where
marginal benefits and costs are equal, this can be seen as an
indicator of the marginal private damage due to the invasive
species. Depending on the species, the private benefits of
control may not be a good measure of the social benefits.
Benefits can be widely dispersed across the different
* Corresponding author. Tel.: þ 44-1904-434072; fax: þ 44-1904432998.
E-mail address: kds2@york.ac.uk (K. Dehnen-Schmutz).
0301-4797/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvman.2003.12.009
stakeholders concerned, and aggregate costs can be difficult
to estimate.
The spread of any invasive species depends on the
control efforts of all those affected. The control costs faced
by each person accordingly depend partly on conditions on
the invaded site, and partly on conditions in neighbouring
sites. Put another way, the private damage costs of harmful
invasive species may be expected to be strictly less than the
social damage costs.
In this paper, we are interested in the determinants of
private expenditure on the control of a particular invasive
species (Rhododendron ponticum), and in the options open
to induce private landowners to undertake socially optimal
level of control. To do this we investigate the extent of the
problem of the invasive non-native plant R. ponticum in the
British Isles, both from an ecological and economic
perspective. We consider the species in its entire synanthropic area thus including a wide spectrum of different
natural and socio-economic conditions. We use a survey of
potentially affected groups to generate data on control costs,
the reasons why people control R. ponticum, and what
control options they consider. Surveys are a standard way of
324
K. Dehnen-Schmutz et al. / Journal of Environmental Management 70 (2004) 323–332
generating such data in the social sciences. They have also
been used by biologists. Perrins et al. (1992), Kowarik and
Schepker (1998) and Williamson (1998) have used questionnaire surveys before to get results on the perception of
non-native species in general and their control in a certain
areas. The taxonomy in the paper follows Stace (1997).
1.1. Rhododendron ponticum as invasive non-native plant
in the British Isles
R. ponticum was introduced into Britain as an ornamental
garden plant about 1763 (Curtis, 1803). Its native area is in
the region of the Black Sea (Turkey, Georgia, Bulgaria)
with disjunct occurrences in Lebanon, Spain and Portugal.
The plants occurring in the British Isles today originate from
Spain mostly (Milne and Abbott, 2000). First reports on its
self-sowing abilities in Britain date back to the year 1849
(Hooker, 1849). Seeding may be one reason for its success
as an ornamental plant in gardens and parks as well as for
the use in extensive plantings for game cover, especially
pheasants, in woodlands, particularly in the 19th Century
(Elliott, 1996). Where conditions favoured establishment, it
subsequently spread from these sites. R. ponticum was also
used almost exclusively as the stock for grafted rhododendrons (Royal Horticulture Society, 1951), but it is so
vigorous that it constantly sends up suckers which, if not
removed, gradually take all the sap from the plant,
eventually causing the grafted variety to die off (Cox,
1998). It is not known how far occurrences of R. ponticum
go back to plantings of these grafted rootstocks. Today R.
ponticum is widespread over the whole British Isles
occurring in 2238 out of 3844 grid cells of (10 km)2
(Preston et al., 2002).
Naturalisation has taken place in natural and near natural
vegetation like acid oak woods, heaths and bogs. R. ponticum
causes ecological modifications such as changes in the species
composition, light reduction, prevention of regeneration of
native shrubs and trees, thus endangering native species (e.g.
Cross, 1975; Fuller and Boorman, 1977; Shaw, 1984;
Thomson et al., 1993; Compton et al., 2002). However,
forestry plantations, both conifer and broadleaf, are also
invaded causing problems for forest management (Brown,
1953; Robinson, 1980; Tabbush and Williamson, 1987;
Edwards et al., 2000). Consequently, the first experiments
on best eradication techniques took place in 1949. They were
carried out by the Forestry Commission (Brown, 1953; Miller,
1954) later followed by control measures in nature reserves
and on private estates (Jones, 1974; Gritten, 1987; Becker,
1988; Searle, 1999; Singleton and Rawlins, 1999).
Today R. ponticum is probably the major alien
environmental weed in the British Isles (Williamson, 2002).
the presence of R. ponticum they will control the species
up to the point where the private marginal benefits of
control (the private marginal damage avoided) is equal to
the private marginal costs of control. But control of R.
ponticum in any one site also reduces the probability that
it will spread in other sites. So the social marginal
benefit of control (the social damage avoided) includes
the damage avoided on contiguous sites. Since landowners in general have no incentive to increase control
effort up to the point where their private control costs
equal the social damage avoided, if landowners are left
to choose the level of control they exercise, there will be
too little control from the perspective of society.
To see the problem consider the following simple model.
Let output on the ith of n contiguous sites, Q i, depend on a
vector of m inputs, Xi ; and on the extent of R. ponticum on
that site, Ri : Further, let the extent of R. ponticum on the ith
site
on the level of control undertaken in all sites,
Pn depend
j
j¼1 E :
That is
0
2
31
n
X
Qi ¼ Qi @xi1 …xim ; Ri 4 Ej 5A
ð1Þ
j¼1
We can identify two problems: a private problem and a
social problem.
The private decision problem takes the form:
MaxX i ;Ei P i ¼ pQi 2 cðXi ; Ei Þ
ð2Þ
subject to Eq. (1), where
Pi
p
ci
profit on the ith site
the price of output
cðXi ; Ei Þ; costs on the ith site.
The first order necessary conditions for the solution of
this problem include the following:
pQiX i 2 ciX i ¼ 0
ð3Þ
pQiRi RiEi 2 ciEi ¼ 0
implying that private landowners will equate the private
marginal benefits and costs of their control effort. The social
decision problem takes the form
MaxS P ¼
n
X
½pQi 2 cðXi ; Ei ðSÞÞ
ð4Þ
i
subject to Eq. (1) and the way in which private effort
depends on the policy instruments, S: The first order
necessary condition for the solution of this problem includes
the following:
1.2. The economic problem
The economic problem investigated is the following.
Since each landowner incurs direct costs from
pQiRi RiEi ESi þ
n
X
j–i
pQjRj RjEi ESi 2 ciEi ESi ¼ 0;
;i
ð5Þ
325
K. Dehnen-Schmutz et al. / Journal of Environmental Management 70 (2004) 323–332
P
In which nj–i pQjRj RjEi is the impact of control on the ith
site on the value of output on all other contiguous sites. In the
private decision problem the decision-maker has no incentive
to take into account the effects of their actions on others.
Solution of the social decision problem requires that they do.
In the absence of intervention, there will be too little control.
The implication of this is that it would be socially desirable to
induce people, through choice of the policy instruments, S; to
undertake the socially optimal level of control.
We have two problems to solve. The first is the estimation
of the marginal external damage costs
P of R. ponticum in any
one site—i.e. the value of the term nj–i pQjRj RjEi : The second
is the specification and estimation of a control effort function
in order to calculate the optimal value of S: We take the general
form of such a control effort function to be Ei ¼ Ei ðZi ; SÞ
where Z is a vector of control inputs.
In other words we wish to calculate (a) how far current
control efforts may fall short of the socially optimal control
effort, and (b) what policy, S; would induce the socially
optimal level of control.
In order to cover all areas where R. ponticum causes
problems four target groups were identified. For problems in
nature conservation these were the managers of nature
reserves and national parks. These include both public
authorities (group 1) and non-governmental organisations
(group 2). The two groups manage the majority of nature
reserves in the UK. In forestry, Forest Enterprise, the
executive agency of the governmental Forestry Commission, is responsible for the management of 8,00,000
hectares of woodland (1/3 of the woodland in Britain).
The Forest Service in Northern Ireland and Coillte, a state
owned company managing 70% of woodlands in the
Republic of Ireland, represent group 3. Private owners of
large estates and forests form group 4. Questionnaires were
sent to groups 1– 3 across the country. Private owners were
asked in the magazines of the Country Land and Business
Association and of the Scottish Landowner Federation to
request a questionnaire if they have R. ponticum on their
estate. Additionally, any information available, e.g. from
respondents of the questionnaire, was used to contact
private estates known to have R. ponticum. In contrast to the
other three groups, preselection excluded nil answers in
group 4. Since private landowners without R. ponticum had
no incentive to take part in the survey, we decided to use the
remaining three groups to obtain information on the absence
of R. ponticum.
In total, 301 questionnaires were sent out, 173 of them
from February to June 2002. From November to December
2002, the group of public authorities was extended to local
governments (County Councils) with 128 additional questionnaires. Table 1 shows the number of recipients and
respondents for all four groups. As some of the recipients
circulated the questionnaire further, the total number of
recipients is unknown and the number of respondents may
be higher than the number of recipients (e.g. in public nature
conservation authorities).
2. Impact and control costs of R. ponticum
2.1. Description of the survey
Data were obtained by designing and administering a
questionnaire to landowners and land managers. The
questionnaire elicited information on three main topics:
(a) general information about R. ponticum and its presence
on sites, (b) the control regime, and (c) the costs (see
Appendix A). Addressees were asked to give estimates if
they did not know the exact numbers requested. The
questions about the control regime and control costs relate
to 2001. Figures for 2000 were also accepted if control work
was not possible in 2001 due to the Foot and Mouth Disease
(FMD) outbreak in the UK in that year. In some areas FMD
prevented movement and work in the countryside for more
than half the year. Recipients were asked also to report
the non-occurrence of R. ponticum in their area of
responsibility.
2.2. General information on sites with R. ponticum presence
In total, the respondents gave information on some
52,000 ha affected by R. ponticum in 248 different sites.
Table 1
Number of questionnaires sent out and total number of responses
Recipient
Total
Public authorities
Nature conservation
Local government
Charities
Forestry
Private landowners
Recipients
Respondents
301
187
38
46
128
33
71
65
33
27
31
16
Completed questionnaires
R. ponticum: yes
Questionnaire: no
R. ponticum: no
105
23
10
46
12
14
41
41
8
15
8
15
9
10
14
1
2
0
The responses are divided according to the type of answer received: completed questionnaires, respondents saying they have R. ponticum on the sites they
are responsible for but did not fill in the questionnaire (‘R. ponticum: yes, questionnaire: no’) and respondents without any R. ponticum (‘R. ponticum: no’).
326
K. Dehnen-Schmutz et al. / Journal of Environmental Management 70 (2004) 323–332
Table 2
Areas (ha) covered by R. ponticum today, 10 years ago, new growth and removed in the last 10 years, and controlled in the reference year 2001 (2000)
Area (ha)
Covered today*
10 years ago
New growth in the last 10 years
Removed in the last 10 years
Controlled in 2001 (2000)
Public authorities
Charities
Forestry
Private landowners
4252.66
359.93
47485.70
771.72
4661.27
655.18
22975.50
211.79
24.50
43.03
3935.50
53.30
505.28
315.20
2586.00
96.47
146.72
103.01
1003.10
21.70
Total
52870.01
28503.74
4056.33
3502.95
1274.53
*Including additional areas of 12120 ha and 606 ha, respectively, mentioned by respondents in forestry and private landowners, for which they were not
able to provide more detailed information in the questionnaire.
This figure includes 12,726 ha affected in forestry for which
no further information was provided. As only some of the
respondents provided information on the total number of
reserves/sites or areas for which they are responsible, this
information was (so far as possible) obtained from other
sources like websites from wildlife trusts. This produced an
estimate of 2678 sites implying that about 9% of all sites are
affected by R. ponticum.
The most affected group is forestry (Table 2). Consequently, the most affected habitat type is woodland (89% of
total area). A more detailed characterisation of the type of
woodland was given only by a few of the respondents. They
mentioned oak woods, broadleaf woods, native pine woodlands and conifer plantations. However, 26,698 ha (75%) of
the affected woodland are registered in the Ancient Woodland Inventory, i.e. land that has had continuous woodland
cover since at least 1600 AD (Kirby et al., 1998). Ancient
woodland, even if it is covered by conifer plantation today,
may still contain ancient woodland species and is regarded
as important for the regeneration of native woodlands (Pryor
et al. 2002). Many respondents did not assign the area
affected to a single habitat type and 869 ha (2.2% of total
area) are therefore described as a mixture of woodland,
heath and moorland. Heath as a single habitat type (117 ha)
is described mostly as lowland heath, but respondents
mentioned also dune heath and heather moor. Other small
scale habitat types are bogs, mires, riverbanks, grassland
and salt marsh.
Most of the respondents gave information on the area
covered by R. ponticum today, but not all of them were able
to answer the questions relating to the change in cover over
the last 10 years.
Of the 248 sites mentioned 186 are nature reserves,
among them 39 National Nature Reserves (NNRs). These
are nature reserves with the highest protection status in the
UK. Table 3 shows the areas affected by R. ponticum in
nature reserves of different protection status. According to
the respondents the main ecological impact of R. ponticum
on the sites concerned was displacement of native species
(193 sites), followed by habitat change (153). On 26 sites
hydrological impacts were reported. This information was
related to the area of R. ponticum at a site and the designated
nature conservation status of the same site. The results
(Table 3) show the ecological impact of R. ponticum
measured in hectares of nature reserve areas. There is no
difference in the perception of the ecological impact
between nature reserves of different conservation status
and sites without designated status. For nearly all sites,
respondents reported displacement of native species and
habitat changes as consequences of R. ponticum invasion. A
comparison of reported hydrological impacts is not possible
as no evidence was given on the hydrological characteristics
of the sites in general.
Respondents for the three impacts mentioned above had to
tick a box per site only. They were also asked to name species
that had become extinct locally due to the invasion of R.
ponticum. This question was answered for 17 sites and the
following species were named: Hyacinthoides non-scripta (3
sites), Eriophorum vaginatum (1 site), Drosera rotundifolia (1
site) and heather (species not specified: Calluna vulgaris
(probably) or Erica spp (possibly)., 5 sites), Quercus spp. (6
sites), Betula spp. (6 sites), and Pinus sylvestris (2 sites).
Other impacts mentioned by the respondents for single
sites included visual effects, overshading of watercourses,
Table 3
Ecological impact of R. ponticum on sites without nature conservation status and in nature reserves
R. ponticum area
Area affected by displacement of native species
Area affected by habitat change
Area affected by hydrological effects
No status
AONB
LNR, NP
SSSI
Part SSSI
NNR
Total
970.5
944.0
911.7
16.0
157.1
141.1
140.1
0.0
3567.4
3551.4
3526.1
3408.3
23062.6
23010.4
17932.3
9057.7
2025.3
2007.3
2021.2
0.0
2122.2
2110.0
2090.9
66.8
31905.0
31764.1
26622.2
12548.8
AONB: Area of Outstanding Natural Beauty; LNR: Local Nature Reserve; NP: National Park; SSSI: Site of Special Scientific Interest, a category that also
includes SAC (Special Area of Conservation), SPA (Special Protection Area) and RAMSAR sites. Part SSSI: only a non-reported percentage of the site has the
status of a SSSI. NNR: National Nature Reserve, the highest protection status in the UK. All figures are given in hectares (ha).
327
K. Dehnen-Schmutz et al. / Journal of Environmental Management 70 (2004) 323–332
impeding of access, security (‘hiding place for muggers’),
damage of fencing, loss of grazing and poisoning of
livestock.
2.3. Control of R. ponticum
In the reference year, 69 respondents of the questionnaire
controlled 1275 ha of Rhododendron (see Table 2). The
average area controlled was 18 ha, but the median is at
2.5 ha, indicating that the majority of respondents controlled
smaller areas. These results are influenced strongly by the
answers of one of the nine respondents from forestry where
from the 1003 ha controlled by this group, 850 ha were in
the Cowal and Trossach Forest District in Scotland alone. In
this Forest District a partly EU funded project took place
in the reference year.
On 121 sites the aim of control measures was the total
eradication, on 59 sites part eradication and on 68 sites
containment. These aims were reached adequately on 133
sites, whereas for 57 sites the respondents came to the
conclusion that the control measures undertaken were not
sufficient to reach their aim.
R. ponticum control was mainly carried out by manual
methods combined with herbicides (47 respondents). Nearly
half of the respondents (22) also used mechanical methods.
The most common disposal method was burning (52); just
15 respondents did not do anything with the removed plants.
Thirty seven respondents controlled with the help of
volunteers. Most volunteers were working for environmental charities, but some public authorities like the Killarney
and Snowdonia National Parks also used volunteers for
clearance works. In total, more than 36,000 volunteer hours
were spent working on control measures.
The total control costs in the reference year were
£530,003. That is an average of £416 per hectare. Again,
the project taking place in the Cowal and Trossach Forest
District has a strong impact on the average values. The
reference year was the last year of a five year project in this
Forest District. In parts of the district control measures
included follow-up spraying of regrown Rhododendron and
monitoring for regrowth and new seedlings. This explains
the relatively low costs of £148 per hectare, the third lowest
value of all respondents. Details on control costs per hectare
given in the literature (Burton and Carpenter, 1999; Searle,
1999; Cross, 2002; Wong et al., 2002) or specified by the
Forestry Commission for grant applications are very
variable depending on the control methods used, the density
and age of the Rhododendron stand and the accessibility of
the sites. They vary widely from £150 for chemical control
of re-grown Rhododendron up to £10,000 as the maximum,
which is assumed to be reached only in exceptional
circumstances like at steep cliff sites accessible for hand
cutting and by rope only. With our results, a detailed
analysis and comparison of the control costs per hectare is
not possible because the questionnaire did not contain
questions about the age structure of R. ponticum stands or
site conditions. In four cases control costs per hectare were
estimated to be more than £10,000 per hectare (£60,000/ha
for one respondent). In general, we assume that respondents
were more likely to be wrong in their estimates of the areas
rather than in the costs they had paid. Costs per hectare were
not calculated by the respondents themselves and may be
biased by false area estimates. Therefore, in the further
analysis of the data (see the control effort model below)
these four exceptional observations were excluded.
Once R. ponticum is eradicated, there may be costs for
the restoration of habitats, e.g. the planting of native plants.
The costs for this specified by eight respondents total
£126,585. Furthermore, R. ponticum could cause costs not
related to its control. Six respondents mentioned additional
costs of £14,336 e.g. for surveying of the sites concerned or
scraping of leaf litter. Adding these sums to the control
costs, the total costs caused by R. ponticum in the reference
year are £670,924.
Most control funds did not come from external
sources. Respondents estimated that 36% of their total
costs were met from external sources (Table 4). Private
landowners and environmental charities were best able to
cover their expenses by external funding (62 and 56%,
respectively), whereas public authorities and forestry
cover less of their costs by external funding (30 and
25%, respectively).
Respondents were asked if they want to control
Rhododendron in future, and if yes, what their financial
needs for that would be per year over the next five years.
The answers of 56 respondents add up to £17 million per
year. Related to the total area managed by the same
respondents that would be £2800 per hectare over the
whole five years. This seems to be a reasonable sum
compared with the figures given in the literature or by
the Forestry Commission. Respondents reporting financial
needs above £10,000 per hectare were excluded from the
analysis. The remaining answers of 41 respondents add
up to £4.5 million per year over the next five years at an
average annual cost of £847/ha. This might be an
underestimate but it should be noted that some of the
respondents might not want to control R. ponticum
completely on their land.
Two respondents did not want to control Rhododendron in their area at all. This is partly because there are
Table 4
External funding sources for R. ponticum control in 2001
Funding source:
£
Forestry grants
EU
Public authorities
Landfill tax
Heritage lottery fund
Other
Total
93215
64500
42920
24000
8200
10500
243335
328
K. Dehnen-Schmutz et al. / Journal of Environmental Management 70 (2004) 323–332
also benefits offered by R. ponticum in terms of aesthetic
appeal, recreational amenity and tourism during the
flowering time in May/June. R. ponticum is an attraction
for tourists especially in regions where it covers large
areas of open land. Addressees were asked if there were
more visitors on their land during that time, and if they
had any revenues from increased entrance fees. Seventeen respondents reported more visitors at the sites, but
none was able to quantify the numbers or additional
revenue. Six respondents specified returns for the sale of
charcoal, chippings and leaves totalling £885 in the
reference year.
2.4. The external costs of R. ponticum control
To identify the best policy response to the problem of
R. ponticum we need first to estimate the external costs
of R. ponticum control. To do this using the data
provided by the survey requires us to make some rather
strong assumptions about the marginal propensity of
R. ponticum to spread from one site to another.
Specifically, we assume that the marginal propensity to
spread is equal to the average propensity to spread and
that this is appropriately measured by the proportion of
sites recording neighbouring sites as the initial source of
R. ponticum.
There are a number of difficulties with this approach,
but it does allow us to distinguish between sites where
R. ponticum was intentionally introduced (either as an
ornamental plant in its own right, or as rootstock for other
cultivars) and sites to which it spread. The survey provided
data on the origin of R. ponticum in a total of 197 sites. It
was originally planted in 117 sites, and was recorded as
having invaded from elsewhere in 102 sites. Note that in 22
sites R. ponticum is due both to original plantings and to
invasion from neighbouring sites. The source of infestation
was unknown at 42 sites, and 1987 sites reported no R.
ponticum at all.
To obtain a probability of spreading we need a true
dimension. The first occurrence (whether planting or
invasion) was only recorded for 84 sites and in many
cases was a rough approximation. The results are
summarised in Table 5 for the 67 sites where respondents
also gave information on the source of infestation. This
subsample implies a cumulative probability that a site will
be invaded rising from 20% in the 19th century, to 38% by
Table 5
Reputed date of planting/first arrival of R. ponticum at 67 sites
Before 1900
1900–1950
After 1950
Planted
Not planted
8
30
1
2
18
8
1950 and 89% by 2000. This compares with 47% for the
full sample (102 of 219 sites were we have the
information).
The cumulative probability of invasion into new sites
indicates that R. ponticum control in any one site reduces the
probability of invasion in neighbouring sites by 47%. That
is, the term
X
X j j
pQRj REi ¼ 20:47p QjRi
j–i
j–i
This is a very rough proxy for the true value of the
R. ponticum externality, but it provides a plausible test for
the optimality of the current social commitment to
R. ponticum control.
Without discriminating between landowners, the survey
reveals (Table 4) that 36% of control costs were met from
external sources—whether forestry grants, EU, to national
public authorities or the national lottery.
Assuming damage costs to be site independent such that
a ¼ pQiRi ¼ pQhRh for all i; h [ {1; 2…n}; the first order
necessary conditions for a social optimum require that social
p
optimal control costs, ciEi ; satisfy
p
ciEi ¼ aRiEi 2 ðn 2 1Þa0:47
and hence that
p
ciEi ¼ ciEi þ ðn 2 1Þa0:47
i.e. that the social benefits of R. ponticum control optima
include both the damage avoided on the ith site and the
marginal external benefits of control, the damage avoided
on n contiguous sites
It follows that, if RiEi ¼ 21 (a unit of control on the ith
site reduces R. ponticum on that site by the same amount),
the ratio of marginal social control costs to marginal private
p
control costs is ciEi =cEi ¼ 1 þ ðn 2 1Þ0:47:
From the survey we know that 64% of control costs are
currently not funded from external public sources implying
p
that ciEi =cEi ¼ 1:57:
This ratio of social to private control costs would satisfy
the conditions for a social optimum if n ¼ 2:2: Put another
way, the current average subsidy is socially efficient
under the various assumptions of the paper if the number
of contiguous sites affected by control in site i were 2.2.
If we can obtain data on the mean number of
contiguous sites, we can estimate the socially optimal
p
ratio ciEi =cEi : A sample of respondents was surveyed
yielding an average of contiguous sites of n ¼ 5:6;
p
implying that ciEi =cEi ¼ 3:16: From that we can calculate
the optimal social control costs c* ¼ £1; 350;628, which
is £679,704 more than the actual funding. Table 6
specifies the results for the different groups of the survey.
Due to the small number of sites in some of the groups
we used the average of n for all groups to calculate the
funding gaps for the groups. The only figure on the
number of contiguous properties available for comparison
329
K. Dehnen-Schmutz et al. / Journal of Environmental Management 70 (2004) 323–332
Table 6
The external costs of R. ponticum control in 2001 in total and for the 4 groups
Total costs
External
Internal
n*
n
Number of sites
Std. Deviation
c*
Difference to actual spending
All
Public
Charities
Private
Forestry
670 924
243 335
427 589
2.2
5.6
32
8.7
1 350 628
679 705
228 098
67 860
160 238
1.9
7.7
9
16.0
506 145
278 047
134 813
76 045
58 768
3.8
3.5
10
1.7
185 630
50 817
60 133
37 430
22 703
4.5
7.6
5
5.2
71 712
11 579
247 880
62 000
185 880
1.7
4.6
8
3.3
587 141
339 261
Total costs (£)-external funded and internal; n* ¼number of contiguous sites affected by control effects in anyone site with the current level of funding;
n ¼ average number of contiguous sites and standard deviation for a sample of sites (number of sites); c* ¼optimal social control costs, i.e. the costs of a
control level sufficient to prevent damage in contiguous sites calculated using the average of n for the whole sample.
is the number of 6.5 (Keeling et al., 2001) derived from
data of all farms in the United Kingdom by tessellation
about each farm location, using the area of each farm as
a weighting. If we used this figure for n it would imply a
p
ratio of ciEi =cEi ¼ 3:59 and a funding gap of £861,982 for
our dataset.
While the socially optimal level of control effort will
generally be site specific, this first order approximation
of the difference between actual and socially optimal
levels of control effort indicates that landowners in each
group should be committing more to R. ponticum control
than they currently do. However, we note that the
funding gap varies significantly between groups.
3. Model of the control effort
As professional labour is included in the control cost
term, volunteer labour is not, but may have influence on the
control effort. Therefore, it is included in the model as a
variable.
Taking logs of both sides, the estimated function can now
be written in the form
log E ¼ a þ b1 log ARp þ b2 log Cost þ b3 log Vol
þ b4 log Fund þ b5 log Rain þ b6 log ChCtrl
Table 7 shows that the model (for a sample size of 62)
explains 56% of the control effort. The values of the
coefficients indicate that control effort is positively and
significantly affected by the area of Rhododendrons,
precipitation, use of chemical methods and external
funding. Control costs have the biggest negative impact
on the control effort as, surprisingly, does volunteer activity.
The latter apparently reflects the fact that volunteers cannot
use cost-effective tools like chainsaws, machinery or
herbicides without violating safety regulations.
The model was also tested with additional parameters
like lowest temperatures, sensitivity against the different
groups and nature conservation status of the controlled sites.
The next step is to identify the best mechanism for
achieving the optimal level of control. To do this we
specify a control effort function. This relates control
effort to (a) the scale of the current problem, (b) the
private control costs, (c) external funding of control
inputs, (d) external labour support (volunteer labour). The
area controlled in the reference year is taken to represent
the control effort. We assume a Cobb-Douglas control
effort function, implying an estimated function of the
form:
Table 7
Coefficients and p-values of the regression analysis of the control effort
model
E ¼ ðARpb1 Costb2 Volb3 Fundb4 Rainb5 ChCtrlb6 Þ
Variable
Coefficient
p-Value
(Constant)
Log ARp
Log ChCtrl
Log Cost
Log Fund
Log Rain
Log Vol
22.895
0.153
0.037
20.584
0.031
1.657
20.028
0.075
0.001
0.019
,0.001
0.003
0.001
0.009
Where:
Control effort, i.e. controlled area in 2001 (ha)
total area of R. ponticum (ha)
control costs per hectare (£)
volunteer labour (hours per hectare area
controlled)
Fund
external funding per hectare (£)
Rain
average annual rainfall per hectare (mm)
ChCtrl use of herbicides in the control process, a
dummy variable.
E
ARp
Cost
Vol
n ¼ 62, adjusted R2 ¼ 0.561. Variables: ARp ¼ total area of R.
ponticum (ha); Chctrl ¼ chemical control, dummy variable; Cost ¼ control
costs per hectare (£); Fund ¼ funding per hectare (£); Rain ¼ average
annual rainfall (mm); Vol ¼ volunteer activity (hours per hectare area
controlled).
330
K. Dehnen-Schmutz et al. / Journal of Environmental Management 70 (2004) 323–332
However, none of these parameters had a significant
influence on the control effort in our dataset.
4. Discussion
R. ponticum has negative effects on nature conservation
and forestry in the British Isles—a fact which respondents to
the survey are highly aware of. R. ponticum imposes high
costs on all groups, and many commit significant resources
to control and restore land invaded by R. ponticum.
Nevertheless, in an analysis designed to test the level of
control effort relative to the social optimum, we find that this
level of effort falls significantly short of what is needed. To
be socially optimal, aggregate social expenditure on
R. ponticum control should be increased by more than
100%, but the funding gap differs between groups. Recall
that the funding gap is the difference between current and
socially optimal external funding. The funding gap in public
sector and forestry implies that external funding should be
increased by 122% and 137% respectively. On the other
hand private landowners and charities—the main beneficiaries of external funding at the moment—should receive 38
and 19% more external funding respectively. These figures
are based on the current funding situation and they may be
biased by the fact that public authorities and predominantly
state owned forestry are unable to use external funding to
the same extent as private landowners and charities.
However, compared with the size of the areas controlled
in 2001 and the areas reported to have R. ponticum, the
calculated funding gaps correlate with the size of the area in
the different groups roughly.
We have also been able to identify—if only in a
preliminary way-the effectiveness of alternative ways of
achieving an increase in control effort. Specifically, we find
that the most effective and reliable method for this will be a
subsidy on control costs. From our estimates of the control
effort elasticities, we calculate that 1% subsidy would
induce a 3.8% increase in control effort. The social cost of
this would be £5300. In order to achieve the same result
through external funding, funding would have to be
increased by 3.6% at an aggregate social cost of £19,080.
Volunteers are widely used in Britain and Ireland for
nature conservation, and so-called "Rhody bashing" is one
of the main activities of volunteers. Our results suggest that
the use of volunteers is not an effective way to increase
control. The attempt to attract more people to volunteer
work therefore does not appear to be a good solution,
although volunteers may well be appropriate in specific
cases. Under the present funding situation they seem to be
the only option, especially for some charities. Groundwork,
a volunteer organisation in Ireland, has organised almost
4000 volunteer weeks of Rhododendron clearance.
Approximately 40% of the infested Killarney oak woods
have been cleared and are being maintained clear by
Groundwork volunteers (Barron, 2000).
Regarding the economic effects, our results are confined
mainly to the costs of control and restoration. Damage caused
by R. ponticum in terms of changes in biodiversity has not been
valued, but may be high considering the ecological effects
described by the respondents. In forestry, the losses in timber
production should be calculated to see if there are benefits for
timber production if R. ponticum is removed in a forest. It is
likely that there are also benefits from R. ponticum for
recreation and in horticulture. Including this entire spectrum of
external costs and benefits would almost certainly increase
the socially optimal level of control. So our results should be
taken to indicate a lower bound.
The historic reasons for planting are certainly no
longer valid today. The private landowners whose
ancestors planted R. ponticum on their estates for
amenity reasons a century ago try to get rid of it now.
In game keeping, shrub cover in woodlands is regarded
as a suitable habitat for pheasants if it is between 30 cm
and 2 m tall. R. ponticum becomes unattractive for this
soon, or requires expensive management. Although
R. ponticum is still available from some nurseries in
Britain it is not a species recommended for planting. Nor
has it been fashionable since the middle of the last
century (Elliott, 1996; Cox, 1998).
R. ponticum is an example of a deliberately introduced
invasive species whose spread was accelerated by human
behaviour. It was introduced to most of the sites mentioned
in this study as an ornamental shrub. Today, the planting of
R. ponticum in woodland for game cover is no longer
recommended because it can become invasive (Robertson,
1992). Other non-native plants like Cotoneaster spp.,
Symphoricarpos albus and Prunus laurocerasus are recommended in its place (Robertson, 1992). The evergreen
shrub P. laurocerasus was mentioned by respondents of the
questionnaire as causing similar problems as R. ponticum in
some areas of Wales and Southwestern England.
P. laurocerasus showed the highest relative increase
between the 1930 – 69 and 1987 – 99 survey of plant
distributions in the British Isles (Preston et al., 2002).
Our data do not allow us to estimate the total impact
of R. ponticum in the British Isles, and hence our results
may be biased. The response rate from managers of
wildlife reserves, including responses from the most
affected areas like North Wales, Southwestern England
and the West Coast of Ireland, indicate that the impacts
on nature reserves are well represented. However, the
relative small number of responses from private landowners and the relative high number of respondents from
forestry who did not fill in the questionnaire but reported
problems in their woodlands, suggest that the extent of
the problem may be under-reported by these two groups.
Nevertheless, we consider that the approach adopted
provides a useful way of testing the optimality of
invasive species control generally, and R. ponticum
control in particular.
K. Dehnen-Schmutz et al. / Journal of Environmental Management 70 (2004) 323–332
331
5. Conclusion
Questionnaire (continued)
The continued spread of R. ponticum imposes high costs
in terms of habitat loss, production losses in forestry and
agriculture in the British Isles. Taking these costs into
account we find that current control levels are below the
socially optimal level. This implies that there should be a
significant increase in funding for R. ponticum control by
central or local government, although the size of the
increase varies between landscapes. We also find that a
subsidy on the control costs would be more effective than an
increase in direct grants.
This finding is qualified by the fact that we are still
unable to offer a complete analysis of the costs and benefits
of control. Missing data include estimates of the value of the
loss of biodiversity in affected habitats, the potential loss of
timber production, and the positive recreational benefits
offered during flowering time. We would not expect
inclusion of these data to affect our general conclusion
that there is underfunding of R. ponticum control. However,
they would affect the size of subsidies required to bring
control up to the socially optimal level.
2
Control
2.1 Aims of control: total eradication, eradication in parts of the site,
preventing further spread (containment), other aims; measures
undertaken sufficient to reach these aims, controlled area in the
reference year
2.2 Control in the reference year: methods (manual, mechanical,
chemical, other), disposal method (burning at the site, chipping
and deposit outside, deposit at the site, remove and deposit, other)
3
Economics of control (all questions related to the reference year)
3.1 Labour: number of man hours (professional/volunteers)
3.2 Control costs: labour costs, equipment and equipment operating
costs, herbicide costs, other
3.3 Additional costs: costs not directly related to control, specify
3.4 Restoration: costs for restoring of habitats following control
3.5 Funding: funding organisation, programme, amount
3.6 Financial needs: if the funds were available would you choose to
remove R. ponticum, how much would you need per year over the
next five years
3.7 Revenues from entrance fees during flowering time, sale of wood,
chippings, seedlings, other
4
Impact on woodland management
4.1 Impact on forest operations (harvest, new plantings, maintenance)
and timber growth, other and costs (tick box: positive, negative,
no effect)
4.2 Economic impact in the reference year: change in timber production
and in woodland management costs per hectare (tick box: benefit,
no change, loss (£0–100, 100 –300, . 300), no estimation possible)
Acknowledgements
We are very grateful to all participants of the inquiry, too
many to name here. We thank particularly John Everitt
(Wildlife Trusts UK), John Harvey (National Trust),
Scottish Natural Heritage, National Trust for Scotland,
Colin Edwards (Forestry Commission), John Cross
(Duchas), the Country Land and Business Association and
the Scottish Landowner Federation who helped to improve
and distribute the questionnaire. Katharina DehnenSchmutz had been supported by a fellowship of the
Deutsche Forschungsgemeinschaft.
Appendix
A.1. Questionnaire
1. General information about sites with R. ponticum presence
1.1 Sites with R. ponticum, name of the estate/forest/reserve,
location, size
1.2 Statuses of the sites: respondents had to tick the appropriate boxes
in a list of different statuses of nature reserves
1.3 Description of R. ponticum occurrences: size of the area covered
today, 10 years ago, removed in the last 10 years, area of natural
spread in the last 10 years, was it planted, did it spread from
neighbouring sites, when did it first arrive/was it planted, invaded
habitat type
1.4 Impact: Displacement of native species, changes of habitat,
hydrological effects, other impacts, local extinction of species as a
result of the invasion
1.5 Visitors: more visitors to the reserves during flowering time of
R. ponticum (additional number if known)
(continued)
There were different versions for the target groups (e.g. no questions related to
woodland management in the questionnaire for nature reserve managers).
References
Barron, C. 2000. Groundwork Rhododendron clearance in Killarney
National Park 1981–2000, a report after 20 years. Unpublished report
for Duchas (Killarney National Park).
Becker, D., 1988. Control and removal of Rhododendron ponticum on
RSPB reserves in England and Wales, The Royal Society for the
Protection of Birds, Sandy, UK.
Brown, J.M.B., 1953. The Rhododendron problem in the woodlands of
southern England. Quarterly Journal of Forestry 47, 239–253.
Burton, D., Carpenter, P., 1999. Life after Rhododendron. Managing
rhododendron in Langley Wood NNR and the New Forest. Enact 7,
11– 14.
Compton, S.G., Key, R.S., Key, R.J.D., 2002. Conserving our little
Galapagos-Lundy, Lundy Cabbage and its beetles. British Wildlife 13,
184– 190.
Cox, K., 1998. Rhododendron and Azaleas. Hamlyn, London.
Cross, J.R., 1975. Biological flora of the British Isles: Rhododendron
ponticum L. Journal of Ecology 63, 345– 364.
Cross, J.R., 2002. The invasion and control of Rhododendron ponticum L.
in native Irish vegetation. In: Kowarik, I., Starfinger, U. (Eds.), In
Biologische Invasionen. Herausforderung zum Handeln?, Technical
University of Berlin, Berlin, pp. 329–337
Curtis, W., 1803. Rhododendron ponticum. Botanical Magazine 16, 650.
Edwards, C., Clay, D.V., Dixon, F.L., 2000. Stem treatment to control
Rhododendron ponticum under woodland canopies. Aspects of Applied
Biology 58, 39 –46.
Elliott, B., 1996. Rhododendrons in British gardens: a short history. In:
Postan, C., (Ed.), In The Rhododendron story: 200 years of Plant
Hunting and Garden cultivation, Royal Horticulture Society, London,
pp. 156–186.
Fuller, R.M., Boorman, L.A., 1977. The spread and development
of Rhododendron ponticum L. on dunes at Winterton, Norfolk,
332
K. Dehnen-Schmutz et al. / Journal of Environmental Management 70 (2004) 323–332
in comparison with invasion by Hippophae rhanmnoides L. at
Saltfleetby, Lincolnshire. Biological Conservation 12, 83–94.
Gritten, R.H., 1987. The spread of Rhododendron ponticum-a national
problem. In: Gritten, R.H., (Ed.), Report of a conference, Snowdonia
National Park, Plas Tan-y-Bwlch, North Wales.
Hooker, J.D., 1849. The Rhododendrons of the Sikkim-Himalaya. Reeve,
Benham and Reeve, London.
Jones, W.I., 1974. A rhododendron eradication trial, Rep. No. Information
Paper No 1, Nature Conservancy Council, UK.
Keeling, M.J., Woolhouse, M.E.J., Shaw, D.J., Matthews, L., ChaseTopping, M., Haydon, D.T., Cornell, S.J., Kappey, J., Wilesmith, J.,
Grenfell, B.T., 2001. Dynamics of the 2001 UK foot and mouth
epidemic:stochastic dispersal in a heterogeneous landscape. Science
294, 813–817.
Kirby, K.J., Reid, C.M., Isaac, D., Thomas, R.C., 1998. The Ancient
Woodland Inventory in England and its Uses. In: Kirby, J.K., Watkins,
C. (Eds.), In The Ecological History of European Forests, CAB
International, Wallingford, pp. 323–336.
Kowarik, I., Schepker, H., 1998. Plant invasions in Northern Germany:
human perception and response. In: Starfinger, U., Edwards, K.,
Kowarik, I., Williamson, M. (Eds.), In Plant Invasions: Ecological
Mechanisms and Human Responses, Backhuys, Leiden, pp. 109–120.
Miller, A.D., 1954. Eradication of Rhododendrons. Journal of the Forestry
Commission 23, 52 –55.
Milne, R.I., Abbott, R.J., 2000. Origin and evolution of invasive naturalized
material of Rhododendron ponticum L. in the British Isles. Molecular
Ecology 9, 541 –556.
Perrings, C., Williamson, M., Dalmazzone, S. (Eds.), 2000. The Economics
of Biological Invasions, Edward Elgar, Cheltenham, p. 249.
Perrins, J., Williamson, M., Fitter, A., 1992. A survey of differing views of
weed classification: implications for regulation of introductions.
Biological Conservation 60, 47–56.
Pimentel, D. (Ed.), 2002. Economic and Environmental Costs of Alien
Plant, Animal, and Microbe Species. CRC Press, Boca Raton, Florida,
USA.
Pimentel, D., Lach, L., Zuniga, R., Morrison, D., 2000. Environmental and
economic costs associated with non-indigenous species in the United
States. BioScience 50, 53–65.
Preston, C.D., Pearman, D.A., Dines, T.D., 2002. New Atlas of the British
and Irish Flora. Oxford University Press, Oxford.
Pryor, S.N., Curtis, T.A., Peterken, G.F., 2002. Restoring plantations on
ancient woodland sites. The Woodland Trust, Grantham. www.thewoodland-trust.org.uk/policy/publicationsmore/Restoration_PAWS_
report1.pdf
Robertson, P.A., 1992. Woodland Management for Pheasants. Forestry
Commission Bulletin 106, 1–18.
Robinson, J.D., 1980. Rhododendron ponticum-a weed of woodlands and
forest plantations seriously affecting management. Proceedings Weed
Control in Forestry Conference 1980, 89–96.
Royal Horticulture Society, 1951. Dictionary of Gardening.
Searle, S., 1999. Controlling Rhododendron at Windsor. Enact 7, 15–16.
Shaw, M.W., 1984. Rhododendron ponticum-ecological reasons for the
success of an alien species in Britain and features that may assist in its
control. Aspects of Applied Biology 5, 231 –242.
Singleton, P., Rawlins, M., 1999. From pest to peat. How rhododendron has
been managed at Roudsea Moss NNR. Enact 7, 17–18.
Stace, C., 1997. New Flora of the British Isles, second ed, Cambridge
University Press, Cambridge.
Tabbush, P.M., Williamson, D.R., 1987. Rhododendron ponticum as a
Forest Weed. Forestry Commission Bulletin 73, 1–7.
Thomson, A.G., Radford, G.L., Norris, D.A., Good, J.E.G., 1993. Factors
Affecting the Distribution and Spread of Rhododendron in North Wales.
Journal of Environmental Management 39, 199–212.
US OTA, 1993. Harmful non-indigenous species in the united states, Office
of Technology Assessment, United States Congress, Washington.
Williamson, M., 1998. Measuring the impact of plant invaders in Britain.
In: Starfinger, U., Edwards, K., Kowarik, I., Williamson, M. (Eds.), In
Plant Invasions: Ecological Mechanisms and Human Responses, pp.
57 –68.
Williamson, M., 2002. Alien plants in the British Isles. In: Pimentel, D.,
(Ed.), In Biological Invasions. Economic and Environmental Costs of
Alien Plant, Animal, and Microbe Species, pp. 91–112.
Wong, J., Youde, E., Dickinson, B., Hale, M., 2002. Report of the
Rhododendron feasibility study, School of Agricultural and Forest
Sciences, School of Agricultural and Forest Sciences, University of
Wales, Bangor, http://www.wildresources.co.uk/reports/rhodo_eng.
pdf.