Content uploaded by Daan Bos
Author content
All content in this area was uploaded by Daan Bos on Oct 12, 2017
Content may be subject to copyright.
1
A historical perspective on the effects of trapping and controlling the muskrat (Ondatra 1
zibethicus) in The Netherlands. 2
3
Running title: History of muskrat trapping in The Netherlands 4
5
Author accepted version Pest Management Science doi: 10.1002/ps.4270 6
7
E. Emiel van Loon a, Daan Bos b, c , Caspara.J. van Hellenberg Hubar d & Ron C. Ydenberg d, e
8
a Computational Geo-Ecology group, University of Amsterdam, Science Park 904, 1098 XH 9
Amsterdam. E.E.vanLoon@uva.nl.
10
b Altenburg & Wymenga ecological consultants. P.O. Box 32, 9269 ZR Veenwouden, The 11
Netherlands. d.bos@altwym.nl. 12
c Community and Conservation Ecology Centre for Ecological and Evolutionary Studies, University 13
of Groningen, P.O. Box 11103, 9700 CC Groningen; 14
d Wageningen University, Resource Ecology Group, P.O. Box 47, 6700 AA, Wageningen, The 15
Netherlands. 16
e Centre for Wildlife Ecology, Department of Biological Sciences, Simon Fraser University 17
Burnaby, BC. CANADA V5A 1S6 ydenberg@sfu.ca. 18
19
phone: 00-31-(0)511 474764, fax number: 00-31-(0)511 472740. 20
21
Abstract 22
23
BACKGROUND: 24
Muskrat are considered a pest species in The Netherlands, and a year-round control programme is in 25
effect. We aimed to evaluate the effectiveness of this programme using historical data on catch and 26
effort collected at a provincial scale. 27
28
RESULTS: The development of the catch differed between provinces, depending on the year of 29
colonisation by muskrat and the investment of effort (measured as field hours). The catch did not peak 30
2
in the same year for the various provinces, and provinces that were colonized earlier in time took 1
longer to attain the peak catch. Trapping resulted in declining populations, but only after a certain 2
threshold of annual effort in trapping had been surpassed. On average populations were observed to 3
decline when the annual effort exceeded 1.4 field hour per km of waterway for several successive 4
years. After having reached a phase of greater control, control organizations tended to reduce effort. 5
6
CONCLUSION: We conclude that control measures can make muskrat populations decline, provided 7
that the effort is commensurate with the population size. Our study emphasizes that experimentation is 8
needed to confirm the causality of the findings, to establish the relation with damage or safety risk and 9
to derive an optimal control strategy. 10
11
Key words: historical data, muskrat, pest-species, population dynamics, trapping intensity 12
13
1 INTRODUCTION 14
Ecology has a long history of investigating harvest, driven mainly by its importance to commercial 15
fisheries. 1 The true population of any harvested species is rarely known. It must be inferred from basic 16
knowledge of population dynamics, the natural history of the harvested species, and records of the 17
catch and effort. In general, populations increase when the harvest effort is low, and decline when 18
effort is too high. 2 The absolute number of animals captured in the long term is highest at intermediate 19
levels of trapping effort, 3 namely the point at which the absolute growth rate of the population is 20
highest. This is a desirable aim for a 21
70 years much effort has been devoted to determining the level of fishing that will generate MSY. 22
23
However, there are other situations in which MSY is not the goal, 4,5 or in which other desirable 24
outcomes (conserving biodiversity, reducing bycatch) conflict with MSY. In the case of slowly-growing 25
organisms, the long term economic gain obtained from overharvesting and investing the proceeds can 26
exceed that obtained from sustainable harvest. Clearly undesirable when applied to whales or old 27
growth forests, overharvesting to reduce populations and thus lowering expenditures in the long run 28
can be desirable when applied to pest and invasive species. 29
30
3
The muskrat (Ondatra zibethicus L.) is native to North America and is considered an exotic species in 1
Europe. It was first recorded in The Netherlands in 1941, evidently having spread from central Europe 2
where it had been introduced as a furbearer. Basic reviews of its natural history and ecology are given 3
by Perry 6, Boutin & Birkenholz 7 and Heidecke. 8 The history and result of muskrat introductions in 4
Europe, as well as their dispersal rates and the impact of muskrat on biota and their habitats in north-5
western Europe, are discussed in Danell. 9 Nowadays, muskrat are present everywhere in the 6
lowlands of north-western Europe, 10,11 and in some regions a control programme is in place. With how 7
much conviction and by what strategy the control is implemented, however, varies greatly by region. 8
9
Muskrat have high reproductive potential. A pair produces on average three litters of approximately six 10
young. 12 Mortality is high, especially in fall and winter. Population trajectories show great seasonal 11
fluctuations 13,14 and there is also evidence for regular annual cycles on the North American continent. 12
1518 Muskrat populations are sensitive to extreme winter coldness and extreme variations in water 13
levels (droughts and floods 13,19 ). Other factors influencing year-to year variation in population levels 14
include disease, predation, and food abundance. In the absence of harvest by man the densities may 15
become high 9,13,14 with maxima varying by orders of magnitude between habitat and years. Although 16
muskrat are generally site-faithful, a varying proportion of young muskrat disperses from its natal site 17
to settle at distances of several hundreds of meters or even multiple kilometres. 2023 Natality, mortality 18
and dispersal are all affected by population density and show strong seasonal variation. 13,14,24 19
However, these mechanisms are not necessarily straightforward: in some years, muskrat appear to 20
tolerate much higher densities than in other years. 19 21
22
Muskrat are semi-aquatic. The habitat available in The Netherlands is of high quality, as it offers a vast 23
network of waterways, an ideal vegetation, a mild climate, and carefully controlled water levels. 24
Consequently, their populations can grow very quickly. Muskrat readily burrow into river banks, dikes 25
and dams, so threatening the integrity of these structures, which in The Netherlands and other low-26
lying parts of north-western Europe are essential for public safety. 27
28
The arrival of muskrat in The Netherlands (1941) was anticipated for more than a decade. Laws 29
banning the ownership or transportation of muskrat were passed and control programmes organized, 30
4
so that measures could begin immediately the first animals were recorded. The history of the control 1
programme is described by Barends 25, van Koersveld 26 and van Troostwijk. 27 Run initially by the 2
national Plant Disease Service, detailed records were kept from the very beginning on the amount of 3
trapping effort (man hours) and the numbers of animals killed (control programmes utilized lethal traps 4
only; no poison was used; details are described by Plug 28 and Barends 29). Responsibility for the 5
programme was later (1986) passed to the Provinces, who in turn quickly passed its administration to 6
the Dutch water authorities. These have divided The Netherlands into eight regions, each with their 7
own muskrat control organizations. 8
9
The impetus for these control measures was to help maintain the physical integrity of the extensive 10
system of dykes in The Netherlands. It is assumed that control measures lead to lower population size 11
and less damage to the dike system. Generally, population models predict that harvested populations 12
have lower average densities than unharvested ones. 30 But for muskrat no rigorous field studies have 13
been conducted. Errington 19 studied fur refuges in the American state Iowa, and found that muskrat 14
density within refuges is generally higher than outside. Parker & Maxwell 31 report on an experiment 15
with controlled harvesting in different seasons and show that combined harvesting in spring and 16
autumn leads to stronger effects on the muskrat population than harvesting in either autumn or spring 17
alone. There are doubts about the effectiveness of muskrat control in Germany. 32 18
19
Despite this lack of evidence, many authors have expressed the view that the dangers of muskrat in 20
The Netherlands are so obvious that the necessity of intensive trapping requires no discussion. 2527,33
21
35 But the control measures are expensive, large numbers of animals are killed and other species are 22
killed as well, directly or indirectly, as side-effects of the control measures. Hence, there is ongoing 23
public debate within The Netherlands on the desirability and effectiveness of these control measures. 24
36 T25
37, which 26
that predators do not catch die of starvation. Hence, it is possible that trapping effort would have no 27
substantial population effects, because the animals are doomed in any case. Further, there may be 28
alternatives to trapping that could mitigate or prevent damage. Finally, one can imagine scenarios in 29
which an invasive population, such as that of the muskrat, would decline over time regardless of 30
5
trapping effort, due to changes in the predator community, 38 vegetation (c.f. Danell 39), or disease. 37 1
For all these s effectiveness is warranted. 2
3
The main aim of our analysis was a quantitative evaluation of how effective the control program has 4
been at reducing or reversing muskrat population growth. To do so we have assembled data on the 5
history of muskrat catch and control in The Netherlands from 1941-2013. The priority for and hence 6
the budget allocated to control has varied over years and between the various authorities, creating 7
spatio-temporal variation in trapping effort, which allows for an evaluation of the relation between effort 8
and catch. Our primary question was therefore: Is the (relative) change in catches dependent on 9
effort? Secondary questions were: To what extent can variation in catch be attributed to differences 10
among provinces, fluctuations in winter coldness or a regular population cycle? Is the effort required to 11
maintain control lower that that required to gain control? 12
13
2. METHODS 14
2.1 Data collection 15
Data were assembled from annual reports published by the muskrat control organizations in The 16
Netherlands. These reports detail the management organization, the numbers of muskrat trapped, and 17
the effort (field hours) required to capture them. Data were available for the entire time series (1941-18
2013) for almost the entire country. Due to the ongoing changes in organization, the structure and 19
detail available in the annual reports differed somewhat over the years. To help interpret these data, 20
we interviewed past and present staff members, including trappers as well as the first coordinator of 21
the national control programme. We aggregated data province by province, and used the 12 time 22
series to investigate the effectiveness of the control programmes. 23
24
The data were incomplete and of variable quality. Some years had missing values for effort, because it 25
could not be reliably estimated. For other years, assumptions had to be made to express effort in 26
identical units. Prior to 1988, for example, field time was sometimes reported as the number of field 27
staff. We converted this to field hours based on known values of field hours per staff member in other 28
years within the same time frame. Trapping was done both by professional trappers, enlisted by the 29
organizations in charge of the muskrat control, and by bounty hunters. The latter did not report their 30
6
effort, so we assumed that the time required to capture an individual muskrat was on average equal to 1
that of professional trappers. A detailed specification of assumptions is given in the Supporting 2
Information. 3
4
For each province, the amount of muskrat habitat was expressed as kilometres of waterway, 5
estimated for each province as the sum of: 6
1 the length of linear waterways that carry water during more than 3 months of the year; 7
2 double the length of linear waterways that are wider than 6 m, and that cannot be crossed 8
on foot (deeper than 1 m); 9
3 the circumference of lakes and ponds. 10
The data were derived using Geographical Information System (GIS) maps for each province. 40 11
Winter coldness in each year was given as the Hellman figure, the positive sum of all daily mean 12
temperatures below 0°C between 1 November and 31 March (http://www.knmi.nl). 13
14
2.2 Analyses 15
Trapping effort was expressed as field time per kilometre waterway per year (h/km/y), and catch as 16
the number of muskrat caught per kilometre waterway per year (n/km/y). For each province, the time 17
series was divided into five control , based on a classification developed by the Association of 18
Regional Water Authorities (Table 1). This 19
(defined as the year with the highest catch) and three post-peak phases, based on the catch (n/km/y). 20
-. After the catch in all cases 21
rose, eventually reaching a peak that in 11 of 12 provinces exceeded 1.0 n/km/y. Years with catch 22
exceeding 0.35 n/km/y were 23
were , and those with catch less than 0.15 n/km/y were deemed 24
latter situation. 25
26
The catch and effort were summarized per province and per phase of control. Differences between 27
provinces in the average annual effort in the various control phases after the peak were assessed 28
using a linear mixed-effects model with the effort (response) as a function of catch (predictor), with 29
province as a random effect. Effort was considered as a response variable in this case because it is a 30
7
management decision to invest time in response to changing catches. Subsequently, possible 1
relations between the duration of different phases of control were investigated. The degree of cyclicity 2
in the time-series was assessed using visual inspection of periodograms and autocorrelation was 3
evaluated via correlograms. 41 4
5
The relative change in catch between successive years was calculated by subtracting the catch in 6
year (i) from that in the following year (i +1) and then dividing by the catch in year (i). The relation 7
between relative change in catch and effort was evaluated in a series of linear mixed-effect models, 8
considering winter coldness as a possible co-variate, and province as random factor. In total we 9
evaluated four models: 1) a null model, with only province as a random intercept; 2) a model with effort 10
as predictor in addition to the random intercept per province; 3) a model with effort as predictor, but 11
with both random intercept and slope per province, and 4) a model with both effort and winter 12
coldness in addition to the random intercept per province. Models were assessed using their AICc 13
values. All analyses were performed in R 42 using package lme4. 43 14
15
3. RESULTS 16
3.1 Initial invasion, and growth of the control effort 17
The number of bounty hunters reached its maximum of 300 individuals in 1983, and declined to zero 18
by 1992 as the national and provincial bounty systems were abolished (Figure 1). Many former bounty 19
hunters were later employed in the professional service operated by the State and the Water 20
authorities. In our interviews bounty hunters reported that they did not change their trapping strategies 21
under their new labour conditions. Catch and effort remained low until 1961, after which both 22
increased rapidly from hundreds to thousands, and tens of thousands after 1966. At its peak in 1991 23
more than 430.000 catches were made by 431 trappers, a number that further increased to over 450 24
in 2004. The catch declined steeply after 2004, while effort remained approximately constant. 25
26
3.2 Differences in developments between provinces 27
The southern provinces were colonized first (Figure 2). After initial invasion, muskrat populations 28
expanded rapidly in the different provinces, and the control status went through successive phases 29
(classified in Table 1). However, the progression showed great variation between provinces. For 30
8
example, the province Noord-Brabant reached its peak in 1978, but Overijssel and Noord-Holland did 1
not until 2005. The structure that did become apparent in the data was a relation between time of 2
invasion and time to reach peak catch per province: provinces that were invaded earlier took longer to 3
reach their peak catch (Figure 3). The catch peaked at an average level of 2.1 n/km/y (SD = 0.98, n = 4
12), but was higher in Zuid-Holland and Utrecht, and lower in Noord-Holland, Overijssel, Noord-5
Brabant and Drenthe (Table 2). 6
7
There were no significant correlations between the year of invasion, peak year, or the year in which 8
sufficient control was attained. Neither was there any apparent spatial pattern in timing of the peak 9
and the timing of attaining sufficient control that suggested the operation of some common external 10
factor. For example, the province Friesland showed a strong decline in catch after 1994, while in the 11
neighbouring province of Groningen the catch fluctuated around a high level until 2012 (see 12
Supporting Information). At neither the provincial nor the national levels was there any sign of a 13
dominant frequency in the periodograms or cyclicity in the autocorrelation, pointing at the presence of 14
a regular population cycle. 15
16
Only four of the provinces attained the practical management objective of fully under (< 0.15 17
n/km/y) by 2013. The duration 18
n/km/y <0.35) was on average 16.9 years (SD = 10, n = 9), and also differed greatly between 19
provinces. Fluctuations in the catch were prominent in some provinces but not in others (see 20
Supporting Information for more details). In all provinces, the control phases following the peak year 21
were characterized by higher average annual effort (Table 2) than before the peak. The control 22
organizations tended to invest less effort with a declining catch (fixed effects-part: effort = 0.63 + 23
0.12catch, p < 0.0001, conditional R2 of 0.69; with random and normally distributed residuals). 24
25
3.3 Catch and effort 26
Trapping effort significantly affected the relative change in catch (Figure 4). The model involving only 27
effort as a predictor and province as a random effect was best supported by the data (Table 3), with a 28
marginal R2 of 0.29 and a conditional R2 of 0.36. Also the residuals for this model appeared to be 29
random and normally distributed. Overall the relative change in catch decreased with -0.295 (95% CI: 30
9
-0.34 to -0.24) per hour increase in effort (p<0.000, with a marginal R2: 0.29 and a conditional R2: 1
0.36). On average the catch was observed to decline when the annual effort exceeded 1.4 h/km/y. 2
The y-intercept (i.e. the relative change at zero trapping effort) had a value of 0.42 (95% CI: 0.33 to 3
0.51) indicating the net population change without trapping. Three provinces had intercepts that were 4
significantly different from this overall mean value (p < 0.05): Zeeland and Noord-Holland had lower 5
intercepts (-0.13 and -0.12 respectively), while the intercept for Utrecht was higher (+0.10). 6
7
4. DISCUSSION 8
9
4.1 Trapping affects the population size 10
Our main result is that the relative change in muskrat catch is significantly reduced with increased 11
trapping effort, strongly suggesting that trapping affects population size. Prior to the peak year, all the 12
provinces showed increasing catches, in some cases lasting decades, in spite of generally increasing 13
effort. Our interpretation is that under these circumstances, field time limited the catch and effort was 14
not intensive enough to cause a decline. After the peak, catch was limited by muskrat population size, 15
and extra effort further depressed the population, reducing the catch in the following year. 16
17
Currently, from approximately 2004 to 2013, there was a considerable decline and low catches in spite 18
of high trapping effort generally point at low population sizes in The Netherlands. Experience from 19
abroad and from within the country suggests that further decline is possible. In Friesland the catch 20
diminished from 2.4 to 0.1 n/km/y, which is less than half of the Dutch average in 2013. In Flanders 21
(Belgium) the catch also declined almost certainly due to trapping, from well over 42,000 in 2001 (> 22
1.9 n/km/y, even without including data from catches by other parties) to 730 in 2013 (0.03 n/km/y; 23
pers. comm. M. Van der Weeën, Vlaamse Milieumaatschappij). In the UK an entire feral muskrat 24
muskrats. 44 25
26
The data have a few shortcomings that should be recognized. The level of effort inferred from the data 27
, and the trapping result 28
(catch) is prone to reporting error. These inaccuracies are however assumed to be of minor 29
importance relative to the large differences reported in space and time for both variables. From our 30
10
interviews it appears that the dataset as a whole, and our conclusions, are sufficiently robust with 1
respect to these sources of error. 2
3
4.2. Differences between provinces and other sources of variation 4
The slope of the relation between relative catch and effort did not vary between the provinces. In three 5
provinces different levels of effort were required to maintain a stable population size or to make 6
populations decline, in comparison to the other 9 provinces. More effort was required in Utrecht and 7
less in Zeeland and Noord-Holland. In addition, we have observed variation between provinces in the 8
time of (initial) colonization, the year of peak muskrat numbers, and the year and size of peak catches. 9
10
The differences between provinces can be attributed to variation over time in the presence and 11
population density of muskrats in neighbouring provinces and countries. We infer that this, in 12
combination with geography, has greatly affected immigration rates over time. The province of Noord-13
Brabant was the first to be colonized by muskrat, 25 because it was close to sources on the other side 14
of the border with Belgium, where muskrat had appeared earlier. The northern part of the province of 15
Noord-Holland on the other hand, has always been quite isolated and was colonized much later. 16
Provinces also differed in habitat suitability for muskrat. Some provinces had much more suitable 17
habitat or greater quantities. Utrecht and Zuid-Holland had the greatest density of waterways. In 18
addition, a waterway in the low lying peat meadows or peat moors of Utrecht and Zuid-Holland may 19
have supported greater density of muskrat than other landscapes, consistent with previous findings by 20
Bos et al. 45 21
22
Further variation may be due to density-dependent factors. For example, both the ease with which 23
animals can be captured (i.e. catch per unit effort) as well as population growth rate likely vary with 24
population density, perhaps non-linearly. The exact nature of the relationship is highly relevant from an 25
economic point of view and deserves further elucidation. It seems progressively cheaper to maintain 26
control at lower population density. This is corroborated by our finding that in practice, lower 27
investments were made as each new phase of control was attained. Knowledge of the relationship 28
between costs and population size is a prerequisite for the proper calculation of an optimal control 29
strategy. 4 This will require experimentation. 30
11
1
Our 12 time series showed that the catch changed markedly when responsibility for the control 2
programme passed from one organization to another. Such delegation of responsibility often involved 3
a change in management procedures. We identified 24 such management changes, of which three 4
apparently led to a situation of diminished control, while in 10 of these cases there were clear 5
indications that the change in management was directly followed by a situation of greater control. 6
There was no change following 11 of these cases. The control status generally increased when the 7
8
9
importance of the quality of management and quantity of trapping effort. We noted great variation in 10
skill and motivation between individual muskrat trappers, and feel that such differences may be 11
partially attributable to details of the organization and its management, such as the extent to which 12
individual trappers were supported to arrive at a coherent control strategy. Further analysis of such 13
differences in the quality of management and how these may have played a role in the Dutch muskrat 14
control programme are beyond the scope of this paper. 15
16
Why did it take the provinces invaded first longer to reach peak muskrat density? This may to a certain 17
extent be explained by the idea that the control organizations were able to slow down the invasion in 18
the originally-invaded provinces. Provinces invaded later had muskrat coming from multiple directions 19
and in higher quantities, necessitating a quicker response in the investments of effort. They may also 20
have learned from developments elsewhere in the country. 21
22
4.3 Possible confounding factors 23
The changes we have documented here have taken place over recent decades, but they are not the 24
only changes that are potentially important to the population dynamics of muskrats. Though there are 25
no indications of a general change in food availability, or the emergence of disease, the predator 26
community has changed over the years studied. Foxes (Vulpes vulpes L.) have invaded the low-lying 27
provinces, 46 . 47 White tailed sea 28
eagles (Haliaeetus albicilla L.) have settled in several nature areas. 48,49 American mink (Neovison 29
vison S.) are present, since they regularly escape from fur-farms, though no viable population has 30
12
established. 50 Hence it is in principle possible that this factor could explain the overall decline of the 1
muskrat in The Netherlands over recent decades. However, some areas in the country still have high 2
numbers of catches (e.g. the province of Groningen) or indications of high population size (the nature 3
reserve Oostvaardersplassen), even in the presence of all or most of these predators, and Bos & 4
Ydenberg 3 argue that the role of predation in the population regulation of muskrat in The Netherlands 5
is small in comparison to the effects of trapping. This is in contrast to findings in Poland 38 (see below). 6
It seems that the intense control measures are most likely responsible for the population decline. 7
8
4.4 The value of hunting bag statistics and the need for experimentation 9
Catch statistics have often been used to make inferences about population development. Long-term 10
time series from the North American continent provide evidence for regular cycles in the populations of 11
muskrat, differing regionally in cycle length and amplitude. 1518 There, hunting or trapping may be 12
intense on a local scale, but current management regimes prevent overharvesting. In Poland an 13
analysis of the decline in the hunting bags of muskrat identified American mink predation as one of the 14
most important factors affecting muskrat numbers. 38 The catch and effort data presented in this paper 15
were previously used by Hengeveld 51 and Matis et al. 52,53 to describe the processes of biological 16
invasion, and to explain models estimating population parameters such as birth- and death rates. It 17
would be extremely worthwhile to elaborate their quantitative population models. Belgian and British 44 18
data support our findings that populations can strongly decline due to trapping, while an analysis of the 19
German catch data has led to doubt about the effectiveness of Muskrat control in that country. 32,54,55
20
We recommend assembling the data for these countries to help assess the costs of trapping at 21
different levels of intensity and the differences between strategies and landscapes. 22
23
Our data are consistent with the hypothesis that the control measures affect population density, but 24
the findings are not detailed enough as yet to guide policy. As stressed in the Introduction, the Dutch 25
control programmes originally arose due to concern for the integrity of dykes. Hence, for policy 26
purposes it is essential to establish the relationship between muskrat population density on the one 27
hand, and economical damage or safety risk on the other. It may also be helpful to quantify the 28
publicly acceptable level of damage per region of interest. These gaps in knowledge hamper proper 29
policy making at the moment. As formulated before, 3 the benefits that can be derived from guiding 30
13
expensive control programmes like these with information derived from well-designed field 1
experiments, are likely to outweigh the costs of such research. 2
3
It is common to encounter situations with overharvesting in fisheries, or successful population 4
reduction in pest management, but for muskrats in The Netherlands (and possibly Flanders), we have 5
the unique situation that trapping effort is known and can be manipulated in the future. Such 6
experimentation would lead to better insight in the causality of relationships and more precise models 7
of optimal harvesting. 8
9
CONCLUSION 10
Control measures do have an effect on the muskrat populations, provided that the levels of investment 11
are in adequate proportion to population size. The study emphasizes the need for experimentation to 12
confirm the causality of the findings, to establish the relation with damage or safety risk, and to derive 13
an optimal control strategy. 14
15
ACKNOWLEDGEMENTS 16
We highly appreciate the input from two anonymous reviewers, Fred Barends, Jasja Dekker, Sjef 17
Keustermans, Bob Litjens, Dolf Moerkens and Leo Wijlaars. 18
19
REFERENCES 20
1 Haddon M, Modelling and Quantitative Methods in Fisheries, Second Edition. CRC Press, 21
Boca Raton, Florida (2011). 22
2 Begon M, Harper JL, Townsend CR, Ecology: individuals, populations and communities, 23
Blackwell Scientific Publications, Oxford (1986). 24
3 Bos D, and Ydenberg R, Evaluation of alternative management strategies of muskrat Ondatra 25
zibethicus population control using a population model. Wildlife Biology 17: 143155 (2011). 26
4 Clark CW, Optimal control theory in one dimension, in Mathematical Bioeconomics: The 27
Mathematics of Conservation, ed. by Clark CW, John Wiley & Sons Ltd, New Yersey, p 368 28
(2010). 29
5 Robertson PA, Adriaens T, Lambin X, Mill A, Roy S, Shuttleworth CM and Sutton-Croft M, The 30
14
Large-Scale Removal of Mammalian Invasive Alien Species in Northern Europe. Pest Manage 1
Sci DOI: 10.1002/ps.4224 (2016). 2
6 Perry Jr. HR, Muskrats, in Wild Mammals of North America, ed. by Chapman JA and 3
Feldhamer GA , The Johns Hopkins University Press, Baltimore, pp 282325 (1982). 4
7 Boutin S and Birkenholz DE, Muskrat and Round-tailed Muskrat, in Wild Furbearer 5
Management and Conservation in North America, ed. by Novak M, Baker JA, Obbard ME and 6
Malloch B, Ontario Trappers Association, Toronto, pp 315325 (1987). 7
8 Heidecke D and Seide P, Bisamratte Ondatra zibethicus (L.), in Buch der Hege Band 1 8
Haarwild, ed. by Stubbe P, VEB Deutscher Landwirtschaftsverlag, Berlin (1990). 9
9 Danell K. Introductions of aquatic rodents: Lessons of the muskrat Ondatra zibethicus invasion. 10
Wildlife Biology 2: 213220 (1996). 11
10 Le Louarn H and Quéré J-P, Les rongeurs de France: faunistique et biologie. 3e Edition révue 12
et augmentée, Editions Quae, Versailles (2011). 13
11 Ondatra zibethicus ) and impact on ecosystem. 14
Ekologija 58: 357367 (2012). 15
12 Moens R, Étude bio-écologique du Rat musqué en Belgique. Parasitica 34: 57121 (1978). 16
13 Errington PL, On the Hazards of Overemphasizing Numerical Fluctuations in Studies of Cyclic 17
Phenomena in Muskrat Populations. J Wildl Manage 18: 6690 (1954). 18
14 Clark WR and Kroeker DW, Population-Dynamics of Muskrats in Experimental Marshes at 19
Delta, Manitoba. Can J Zool 71: 16201628 (1993). 20
15 Erb JD, Stenseth N and Boyce MS, Geographic variation in population cycles of Canadian 21
muskrats (Ondatra zibethicus). Can J Zool 78: 10091016 (2000). 22
16 Erb JD, Boyce MS and Stenseth NC, Spatial variation in mink and muskrat interactions in 23
Canada. Oikos 93: 365375 (2001). 24
17 Haydon DT, Stenseth NC, Boyce MS and Greenwood PE, Phase coupling and synchrony in 25
the spatiotemporal dynamics of muskrat and mink populations across Canada. Proc Natl Acad 26
Sci U S A 98: 1314913154 (2001). 27
18 Estay SA, Albornoz AA, Lima M, Boyce MS and Stenseth NC, A Simultaneous Test of 28
Synchrony Causal Factors in Muskrat and Mink Fur Returns at Different Scales across 29
Canada. PLoS One 6: 18 (2011). 30
15
19 Errington PL, Muskrat populations. Iowa State University Press, Ames Iowa, USA (1963). 1
20 Aldous SE, Muskrat trapping on Sand lake National wildlife refuge, South Dakota. J Wildl 2
Manage 11: 7790 (1947). 3
21 Mallach N, Markierungsversuche zur Analyse des Aktionraums und der Ortsbewegungen des 4
Bisams (Ondatra zibethica L.). Anzeiger fur Schadlingskunde Pflanzenschutz Umweltschutz 9: 5
129136 (1971). 6
22 Caley MJ, Dispersal and Inbreeding Avoidance in Muskrats. Anim Behav 35: 12251233 7
(1987). 8
23 Adelberg D, Ausbreitungsstrategien des Bisams Ondatra zibethicus (L. 1766) im Tal der 9
Unteren Hasen / Emsland, Universit
ä
t Osnabr
ü
ck, Osnabr
ü
ck (2008). 10
24 Simpson MR and Boutin S, Muskrat, Ondatra-Zibethicus, Population Responses to Harvest on 11
the Old Crow Flats, Yukon-Territory. Canadian Field Naturalist 103: 420422 (1989). 12
25 Barends F, The Muskrat (Ondatra zibethicus): expansion and control in the Netherlands. Lutra 13
45: 97104 (2002). 14
26 van Koersveld E, De Muskusrat, Ondatra zibethica L., in Nederland en zijn bestrijding. 15
Plantenziektekundige Dienst, Wageningen (1953). 16
27 Doude van Troostwijk WJ, The Musk-rat (Ondatra zibethicus L.) in the Netherlands, its 17
ecological aspects and their consequences for man, Rijksuniversiteit Leiden, Leiden (1976). 18
28 Plug K, Handboek Muskusrattenbestrijding, Min LNV, Den Haag (1988). 19
29 Barends F, Bestrijding van Muskusrat en Beverrat, LCCM and TCM, Den Haag (2007). 20
30 Boyce MS, Sinclair ARE and White GC. Seasonal compensation of predation and harvesting. 21
Oikos 87: 419426 (1999). 22
31 Parker GR and Maxwell JW, An Evaluation of Spring and Autumn Trapping Seasons for 23
Muskrats, Ondatra-Zibethicus, in Eastern Canada. Canadian Field Naturalist 98: 293304 24
(1984). 25
32 Pelz HJ,. Mitt Biol Bundesanst Land-26
Forstwirtsch 317: 219234 (1996). 27
33 Kluyver HN, De Bisamrat. Verslagen en Mededelingen van Plantenziektekundige Dienst te 28
Wageningen 85: 132 (1937). 29
34 Ritzema-Bos J, De Muskusrat, Bisamrat of Ondatra (Fiber zibethicus L.). Tijdschrift over 30
16
Plantenziekten 23: 179 (1917). 1
35 Thijsse JP, De Muskusrat. De Levende Natuur 41: 348349 (1937). 2
36 Zandberg F, de Jong P and Kraaijeveld-Smit F. Muskusrat. Op alternatieve wijze schade 3
voorkomen. Bont voor Dieren, De Faunabescherming,Nederlandse Vereniging tot 4
Bescherming van Dieren, Den Haag (2011). 5
37 Errington PL, Factors Limiting Higher Vertebrate Populations. Science 124: 304307 (1956). 6
38 Brzezinski M, Romanowski J, Zmihorski M and Karpowicz K, Muskrat (Ondatra zibethicus) 7
decline after the expansion of American mink (Neovison vison) in Poland. European Journal of 8
Wildlife Research 56: 341348 (2010). 9
39 Danell K, Short-term plant successions following the colonization of a Northern Swedish lake 10
by the Muskrat, Ondatra zibethica. J Appl Ecol 14: 933947 (1987). 11
40 LCCM, Landelijk jaarverslag 2007 muskus- en beverrattenbestrijding, Landelijke Coördinatie 12
Commissie Muskusrattenbestrijding, Tiel (2008). 13
41 Legendre P and Legendre L, Numerical ecology. Developments in Environmental Modelling, 14
Vol. 24. 3rd ed. Elsevier Science BV, Amsterdam (2012). 15
42 R Core Team, R: A language and environment for statistical computing. https://www.R-16
project.org/ [accessed 2015] 17
43 Bates D, Maechler M, Bolker M and Walker S, lme4: Linear mixed-effects models using Eigen 18
and S4. R package version 1.1-9. https://CRAN.R-project.org/package=lme4 [accessed 2015] 19
44 Gosling LM and Baker SJ, The Eradication of Muskrats and Coypus from Britain. Biol J Linn 20
Soc 38: 3951 (1989). 21
45 Bos D, van Belle J, van Wieren S, Ydenberg RC and Goedhart PW, Naar objectieve schatting 22
van aantallen Muskusratten in Nederland. De Levende Natuur 111: 9499 (2010). 23
46 Mulder JL, Vos Vulpes vulpes, in Atlas van de Nederlandse zoogdieren. Natuur van Nederland 24
12, ed. by Broekhuizen S, Spoelstra K, Thissen JBM, Canters KJ and Buys JC, Naturalis 25
Biodiversity Centre, Leiden, pp. 236238 (2016). 26
47 Bijlsma RG, Hustings F and Camphuysen CJ, Algemene en schaarse vogels van Nederland 27
(Avifauna van Nederland 2), GMB Uitgeverij/KNNV Uitgeverij, Haarlem/Utrecht (2001). 28
48 van Rijn S, Zijlstra M and Bijlsma RG, Wintering White-Tailed Eagles Haliaeetus albicilla in the 29
Netherlands: Aspects of Habitat Scale and Quality. Ardea 98: 373382 (2010). 30
17
49 Boele A, van Bruggen J, Hustings F, Koffijberg K, Vergeer J-W and van der Meij T, 1
Broedvogels in Nederland 2013, Sovon Vogelonderzoek Nederland, Nijmegen (2015). 2
50 Dekker JJA and Hofmeester TR, The status of the American mink (Neovison vison) in the 3
Netherlands. Lutra 57: 515 (2014). 4
51 Hengeveld R, Dynamics of Biological Invasions, Chapman & Hall, London (1989). 5
52 Matis JH and Kiffe TR, Effects of immigration on some stochastic logistic models: A cumulant 6
truncation analysis. Theoretical Population Biology 56: 139161 (1999). 7
53 Matis JH, Kiffe TR and Hengeveld R, Estimating Parameters for Models From Birth-Death-8
Migration Abundance Data: Case of Spatio-Temporal Muskrat Spread in the Netherlands. 9
Journal of Agricultural, Biological and Environmental Statistics 1: 4059 (1996). 10
54 Halle S and Pelz HJ, Zur Effizienz der Bekämpfung des Bisams (Ondatra zibethicus) anhand 11
von Fangdaten aus dem Land Bremen. Zeitschfirt fuer Angewandte Zoologie 77: 204218 12
(1990). 13
55 Pelz HJ, Spread of the Musk-Rat in the Federal Republic of Germany. Bull. OEPP 14: 153157 14
(1984). 15
16
18
Figure legends 1
Figure 1. The number of muskrat trapped (dots, catches/km of waterway/y) and the effort (field 2
hours/km/y) invested in The Netherlands as a whole, 1941-2013. Filled dots and triangles indicate the 3
year with maximum values for the catch or the effort. Totals for each province are presented in the 4
Supporting Information. 5
6
7
Figure 2. Timeline indicating the years when successive phases of muskrat control (defined in Table 8
1) were attained in each of the twelve provinces of The Netherlands. The height of each bar is 9
proportional to the length of waterway in each province. Timelines begin with the year when muskrats 10
were first registered ('invasion') in a province. The inset provides a map with the geographical 11
boundaries per province. 12
19
1
Figure 3. The relationship between the duration of the phase of increase ('invasion to peak’ phase; 2
years) and the numbers of years that had passed after the initial invasion of The Netherlands and the 3
invasion of the province. The line y = -0.85x + 48.8 (p=0.01, R2adj =0.43) represents the linear 4
regression model fitted. The grey area refers to the 95% confidence limits of this model. 5
6
7
Figure 4. The relation between effort (x axis, in field hours per km waterway per year) and the 8
proportional change in the number of muskrats caught. Each point represents a province - year 9
combination (n = 422). Datapoints before (○) and after (+) the peak in a given province are indicated 10
separately. The line y = -0.295x + 0.42 (p < 0.000, marginal R2 = 0.36) represents the fixed part of the 11
20
linear mixed-effects model (i.e. not taking the random intercept per province into account), with the 1
grey area showing the 95% confidence limits. 2
3
4
21
Table 1. Practically-for muskrat management in The Netherlands. 1
Situation
muskrat trapped
(n/km/y)
Before the peak
Pre-invasion
0
Invasion to peak
> 0
After the peak
No control
>0.35
Sufficient control
0.15-0.35
Full control
<0.15
2
22
Table 2. Average annual effort (field hours/km/y) for The Netherlands and per province (acronym in 1
brackets) for the successive phases of control after invasion by muskrat, and the peak catch 2
(number/km/y). No entry in a cell indicates that level of control was not attained. The inset in Figure 2 3
provides a map with province boundaries. 4
Effort per control phase (field hours/km/y)
Province
Invasion
to peak
No
control
Sufficient
control
Full
control
Peak catch
(n/km/y)
The Netherlands
0.6
1.6
1.5
Drenthe (DR)
0.5
1.4
1.2
1.1
Flevoland (FL)a
1.1
1.7
1.5
2.0
Friesland (FR)
0.7
1.4
1.4
1.1
2.5
Gelderland (GD)
0.9
2.1
1.8
2.2
Groningen (GR)
1.4
2.0
2.3
Limburg (LB)
1.0
1.7
1.6
2.1
Noord-Brabant (NB)
0.7
2.0
1.7
1.7
1.7
Noord-Holland (NH)
0.5
1.2
1.0
0.2
Overijssel (OV)
1.4
1.9
1.4
Utrecht (UT)
0.7
2.3
1.8
4.2
Zeeland (ZL)
1.0
1.7
1.6
1.3
2.5
Zuid-Holland (ZH)
0.9
2.3
1.9
3.1
5
a Flevoland was established as a new province on land reclaimed between 1942 and 1968. 6
7
8
23
Table 3. Ranking of models for the relative change in catch that were evaluated according to AICc-1
values. The model-set comprised a null model with only province as a random intercept (null), a 2
model with effort as predictor in addition to the random effect of province (effort), a model with 3
effort as predictor, but with both random intercept and random slope per province (effort-rndslp), 4
and a model with both effort and winter coldness in addition to the random effect of province 5
(effort+cold). K = number of free parameters in the model, AICc = Akaike information criterion, ΔAICc 6
= difference between model AICc and AICc value of the best model, AICcWt =AICc weights. 7
Model:
K
AICc
ΔAICc
AICcWt
effort
4
116.8
0.0
0.82
effort-rndslp
6
119.9
3.0
0.18
effort+cold
5
133.3
16.5
0.00
null
3
1157.7
1040.9
0.00
8
SUPPORTING INFORMATION 9
10
Muskrat catches (dots, catches/km/y) and the field time (triangles, hours/km/y) required to capture 11
them is given per province from 1941-2013. The arrow at the x-axis indicates year of invasion. Filled 12
dots and triangles are the year with maximum values for the catch or the effort. The grey-shaded area 13
indicates the period during which the management was a formal responsibility of the province. Note 14
that in Utrecht there has been an intermediate transfer of management to a public entity in 1996, while 15
in Zuid-Holland the management has been a shared responsibility between province and water 16
authorities as of 1994. 17
24
1
2
25
1
2
3
4
5
6
26
1
2
3
4
27
1
2
3
28
1
2
3
29
1
2
3
4
Overview of assumptions 5
The data are not complete, and of variable quality. In addition to the assumptions specified in the main 6
text we made the following assumptions. 1941: assumed that no field hours have been invested, 7
because the first catch was made by accident and WWII was going on. 1946-49: assumed that a part-8
time trapper spent 600 hour per year. 1978-1982: the field hours of provinces Gelderland and 9
Overijssel have been redistributed over Flevoland, Gelderland and Overijssel in proportion to the 10
number of catches (catches were administered for Flevoland, but field hours were booked upon the 11
two other provinces). 1983-86: assumed that the number of field hours per trapper equalled that in 12
30
1987 = 1370 h/y. 1986-1992: assumed that a bounty hunter spent an equal amount of time trapping as 1
the average other trappers (fulltime and part-time together). 2
3
