Statistical indicators for insects caught in the ... - Wiley Online Library

12 downloads 0 Views 563KB Size Report
Mar 19, 2017 - 2012). Typically, these have investigated changes in emergence date ... Sparks, 2000; Westgarth- Smith, Roy, Scholz, Tucker, & Sumpter,.
|

Received: 13 October 2016    Accepted: 19 March 2017 DOI: 10.1111/jen.12403

ORIGINAL CONTRIBUTION

Statistical indicators for insects caught in the development trap R. P. Blackshaw1

 | P. Esbjerg2

1 Blackshaw Research and Consultancy, Chudleigh, Devon, UK

Abstract

2

Department of Plant and Environmental Sciences, Copenhagen University, Frederiksberg C, Denmark

The effects of climate change on insect phenologies are of current interest, but little

Correspondence Rod P. Blackshaw, Blackshaw Research and Consultancy, New Pound Parade, Chudleigh, Devon, UK. Email: [email protected]

merical simulation modelling to show that increased mortality from a partial second

attention has been given to the potential driver for population decline of increased mortality arising from the failure to reach a viable overwintering stage. We use nugeneration results in the weakening of density dependence, whereas increasing mortality within the first generation does not. These general results are then compared with annual trap count data for the moth Agrotis segetum (Denis and Schiffermüller) in Denmark, a species which is in decline there because a partial second generation fails to survive the winter. Results from the moth data are consistent with those from the simulations, and we conclude that the development trap phenomenon can be detected by the application of simple statistical tests to time-­series data. KEYWORDS

Agrotis segetum, climate change, density dependence, lost generation hypothesis, modelling, time-series data

1 | INTRODUCTION

whereby ectotherms can be driven into population decline as the climate changes, despite a shortening life cycle.

Phenological responses of insects to climate change have been

This is a phenomenon that may afford an additional explanation

the focus of many studies (O’Connor, Selig, Pinsky, & Altermatt,

for some observed population declines, and so there is a question of

2012). Typically, these have investigated changes in emergence date

whether such effects can be detected at the species level using exist-

(Altermatt, 2010a, 2012; Karlsson, 2014; O’Neill et al., 2012; Roy

ing time-­series data. In this article, we take a simple population model

& Sparks, 2000; Westgarth-­Smith, Roy, Scholz, Tucker, & Sumpter,

for a univariate insect, modify it to incorporate increased mortality

2012), synchronization with resources (Altermatt, 2010b; Forrest &

within or between the generations and use numerical simulations to

Thompson, 2011; Hegland, Nielsen, Lazaro, Bjerkness, & Totland, 2009;

investigate whether such mortality effects show systematic patterns

Van Asch & Visser, 2007; Visser & Holleman, 2001) and geographic

that are detectable using simple statistical analyses. These results are

range (Asher, Fox, & Warren, 2011; Menendez et al., 2007; Williams &

then compared with data for a moth species, Agrotis segetum (Denis

Liebhold, 2002).

and Schiffermüller), which is in decline as a result of a development

Some authors have assumed that warming, at least in temperate

trap (Esbjerg & Sigsgaard, 2014).

regions, will enhance population growth, especially if an additional

Agrotis segetum, commonly known as the turnip moth, is a pest

generation is added (Juszczak, Kuchar, lesny, & Olejnik, 2013; Kiritani,

species with a wide distribution in north-­west Europe. Historically

2013; O’Connor et al., 2012). However, in some circumstances, ac-

it was thought of as a problem arising from a single generation (e.g.

celerated development may result in population decline. An import-

see Jones & Jones, 1974) but is increasingly seen to have a second

ant aspect is whether an overwintering stage can be reached before

generation flying in late summer which has been attributed to an ear-

conditions deteriorate (Altermatt, 2010a). This idea was recently for-

lier emergence date and a shortened generation interval (Esbjerg &

malized as the “lost generation hypothesis” leading to the concept of

Sigsgaard, 2014). The soil-­dwelling larvae show an increasing capacity

a development trap (Van Dyck, Bonte, Puls, Gothard, & Maes, 2015)

to cope with adverse conditions as they develop through instars but,

272  |  wileyonlinelibrary.com/journal/jen © 2017 Blackwell Verlag GmbH

J Appl Entomol. 2018;142:272–276.

|

      273

BLACKSHAW and ESBJERG

in Denmark at least, need to reach the sixth instar if they are to survive winter conditions (Esbjerg & Sigsgaard, 2014).

simple univariate insect model, and then show that analyses of changes in moth count levels over time conform to expected outcomes from the numerical simulation modelling. We conclude that there are simple

Mean count

In this study, we demonstrate that the population dynamics of the moth have a density-­dependent component justifying the use of the

statistical indicators that can be used to identify species that may be subject to the development trap effect.

2 | MATERIALS AND METHODS

Sampling year

2.1 | Simulating the timing of mortality We take a density-­dependent model developed to describe annual

F I G U R E   1   Long-­term trends in annual sex pheromone trap catches of Agrotis segetum in Denmark [Colour figure can be viewed at wileyonlinelibrary.com]

changes in populations of the univoltine Tipula paludosa Meig (Blackshaw & Petrovskii, 2007). The generic form of this model is as follows: ) ( Nt ΔNt + 1 = α − β log log Nt

(1)

and in its stochastic form: ( ΔNt = Nt

[ α+r

10

Nt

)

]−β

−1

(2)

In this model, α is the regression intercept and functions as a population-­limiting factor. In other words, the higher the value of α, the more extreme the environmental conditions are and the greater will be within-­generation mortality. β is the regression slope representing the intensity of the density-­dependent feedback, is the long-­term mean population and r is a random variable drawn from a uniform distribution. As this model describes the density-­dependent annual fluctuations in numbers of a univoltine species, it can simulate what happens when additional mortality is introduced either within the generation (year) by altering the value of α to represent mortality from differently limited environments or by applying an additional mortality factor after Nt+1 has been calculated to represent mortality arising from a partial second generation. The T. paludosa model presented by Blackshaw and Petrovskii (2007) had empirical values of −0.822 for β, −0.0252 for α and r ranging between −22.2% and +22.2%. We used these values in numerical simulation models except where indicated below. Three sets of numerical simulations were run: (i) varying α over 5 levels from 0 to −2 with an average population () of

F I G U R E   2   Outcomes from numerical simulations that varied α over five levels from 0 to −2 in Equation 2 to represent increased within-­generation mortality. The data show the mean slopes and R2 values (±SE) from regressions of ΔNt on Nt

0.5 × 106 ha−1, (ii) varying mean population level with  = 0.25, 0.5

Ten replicate simulations were run for each of these consisting of

and 1 × 106 ha−1, applying a partial second-­generation mortality factor

1,000 time steps so that each simulation resulted in 1,000 values of

of 0%–90% by reducing the calculated population at Nt+1 prior to calculating the population at Nt+2, (iii) application of a maximum value for

Nt and 999 of ΔNt. Regression of ΔNt on Nt was performed and plots

made of the relationship between mean slope and mean R2 values

r of ±22.5, 45 or 90% with  = 0.5 × 106 ha−1, applying the partial

against either the values for α (set 1) or percentage entering a second

second-­generation mortality factor of 0%–90%.

generation (sets 2 and 3).

The intention behind this choice of models was to determine whether there were detectable systematic outcomes resulting from varying within-­generation mortality, the proportion of a population

2.2 | Agrotis segetum dynamics

entering a non-­viable second-­generation stage and changing environ-

A full description of the A. segetum regional sampling programme is given

mental variability.

elsewhere (Esbjerg & Sigsgaard, 2014). To summarize: sex pheromone

|

BLACKSHAW and ESBJERG

274      

1

1 0.8

.5

0.6

Slope

0.4

Slope

0.2

0

0

–0.5

–0.2 –0.4

–1

–0.6 –0.8

.4

–1

R2

.3

.4

.2 .1

R2

.3

0

.2 .1 0

0

10 20 30 40 50 60 70 80 90 Percentage entering a second generaon

F I G U R E   3   Outcomes from numerical simulations varying the long-­ term mean population level () in Equation 2. The data show the mean slopes and R2 values from regressions of ΔNt on Nt with a partial second-­generation mortality factor by having different proportions of the population entering a partial second generation. The three lines are  = 0.25 × 106 (dot), 0.5 × 106 (dash) and 1 × 106 (solid) ha−1 [Colour figure can be viewed at wileyonlinelibrary.com]

traps were operated at sites in Denmark each summer yielding a total number of captured adult males in the first generation. The trapping

0

10

20

30

40

50

60

70

80

Percentage entering a second generation

90

F I G U R E   4   Outcomes from numerical simulations varying the stochastic element (r) in Equation 2. The data show the mean slopes and R2 values from regressions of ΔNt on Nt with additional mortality arising from different proportions of the population entering a partial second generation. The three lines are rmax = 22.5% (dot), 45% (dash) and 90% (solid) the next (ΔNt in Equation 2). The subsequent data were then tested for significance using log-­log linear regression (Blackshaw & Petrovskii, 2007) to test whether density dependence is evident for A. segetum. This exercise was undertaken separately for the two time periods 1981–1991 and 1997–2009 and the combined data.

3 | RESULTS There were no significant effects of varying α in simulations on ei-

ther the slope or R2 value of the relationship between ΔNt and Nt

data we use in this study spanned two periods: 1981–1991 and 1997–

(Figure 2). Both the slope and R2 values changed with the percentage

2009. Sites differed between these two periods with seven sampled in

entering a partial second generation, but this was not dependent on

the earlier study compared with 20 from 1997 onwards. Not all sites

either the value of (Figure 3) or r (Figure 4).

were sampled each year. Mean annual counts showed a decline over time as well as a reduction in amplitude (Figure 1). The analytical method used for T. paludosa (Blackshaw & Petrovskii, 2007) was applied to the A. segetum sex pheromone trap counts. Data

The regressions of log(per capita increase + 1) against log(relative abundance) for all A. segetum data and the two time series are shown in Figure 5. The relationship was weaker for the 1997–2009 data than for 1981–1991 period.

points were included if they satisfied the condition of being able to calculate a population change from 1 year to the next. For each sampling site within the two time periods 1981–1991 and 1997–2009

4 | DISCUSSION

meeting this condition, the mean of the site time series was calculated. This provided the value for in Equation 2. A per capita increase

In addition to the long-­term decline in numbers shown in Figure 1,

was then calculated from the difference in trap counts from 1 year to

we have demonstrated that there is density-­dependent feedback

|

      275

BLACKSHAW and ESBJERG

Our numerical simulations show that systematic changes in the relationship between ΔNt and Nt are consistent with mortality arising from a partial second generation that does not survive the winter (Figures 3 and 4). Both the slope of the regression and its strength decline as the proportion entering the second generation increases. This effect is not observed when additional mortality is applied only within the first generation (Figure 2). The data for A. segetum follow a similar pattern with a decline in mean annual population size (Figure 1) and a weakening of the regression slope and its statistical significance (Figure 5). These observations are consistent with the hypothesis that the decline in observed numbers of this pest arises from increasing numbers entering a second (lethal) generation. Agrotis segetum has declined in importance as a pest as the surveys were started in 1981 to the point at which Danish farmers hardly anticipate problems (Esbjerg & Sigsgaard, 2014). There may come a point when sufficient change in the phenology has occurred to allow a second generation to reach the sixth instar and so survive the winter, becoming a serious pest once more. Interestingly, there is a suggestion from the simulation modelling that the density-­dependent feedback could switch from negative to positive at percentages >80% entering the second generation, and so there may be the potential for sporadic outbreaks in advance of wider pest resurgence. The ability to reach a secure developmental stage for successful overwintering has only recently been recognized as a potential driver in the decline of some species. The bigger question is whether these findings have relevance to the reported declines of other Lepidoptera (e.g. Fox, 2013) and insects more generally. Where long time-­series data sets exist for other species in decline, the approach we have developed here could be applied to seek evidence for systematic changes in the relationship between ΔNt and Nt. We suggest that reduction in the strength and significance of this regression over time is indicative of increased overwintering mortality. It is not, however, conclusive by itself and needs to be interpreted through understanding the biology and ecology of the species. Nevertheless, it is a tool that can help better focus research to aid both F I G U R E   5   Plots of log(per capita increase + 1) against log(relative abundance) providing evidence of density-­dependent feedback and conformity to the general model: (a) all data, (b) 1981 to 1991 and (c) 1997 to 2009

conservation and pest management.

AU T HO R CO NT R I B U T I O N PE collected the data. RPB analysed the data and undertook the simulation modelling. Both authors contributed to writing the paper.

operating on A. segetum population dynamics (Figure 5) in a similar way to that shown for T. paludosa (Blackshaw & Petrovskii, 2007), legitimizing our modelling approach. The average strength of the density-­dependent feedback for A. segetum is less than that previously reported for T. paludosa, regression slopes of −0.495 (Figure 5a) and −0.822 (Blackshaw & Petrovskii, 2007), respectively. The strength of the feedback relationship declined from the earlier trapping period (Figure 5b) to the later one (Figure 5c). The question that we ask in this study is whether this provides evidence of changes in the relationship that can be linked to the decline in numbers evident from Figure 1.

REFERENCES Altermatt, F. (2010a). Climatic warming increases voltinism in European butterflies and moths. Proceedings of the Royal Society of London B: Biological Sciences, 277, 1281–1287. Altermatt, F. (2010b). Tell me what you eat and I’ll tell you when you fly: Diet can predict phenological changes in response to climate change. Ecology Letters, 13, 1475–1484. Altermatt, F. (2012). Temperature-­related shifts in butterfly phenology depend on the habitat. Global Change Biology, 18, 2429–2438. Asher, J., Fox, R., & Warren, M. S. (2011). British butterfly distributions and the 2010 target. Journal of Insect Conservation, 15, 291–299.

|

BLACKSHAW and ESBJERG

276      

Blackshaw, R. P., & Petrovskii, S. V. (2007). Limitation and regulation of ecological populations: A meta-­analysis of Tipula paludosa field data. Mathematical Modelling of Natural Phenomena, 2, 46–62. Esbjerg, P., & Sigsgaard, L. (2014). Phenology and pest status of Agriotes segetum in a changing climate. Crop Protection, 62, 64–71. Forrest, J. R. K., & Thompson, J. D. (2011). An examination of synchrony between insect emergence and flowering in Rocky Mountain meadows. Ecological Monographs, 81, 469–491. Fox, R. (2013). The decline of moths in Great Britain: A review of possible causes. Insect Conservation and Diversity, 6, 5–19. Hegland, S. J., Nielsen, A., Lazaro, A., Bjerkness, A. L., & Totland, O. (2009). How does climate warming affect plant-­pollinator interactions? Ecology Letters, 12, 184–195. Jones, F. G. W., & Jones, M. G. (1974). Pest of field crops, 2nd ed. (p. 106). London, Beccles and Colchester: William Clowes & Sons Ltd. Juszczak, R., Kuchar, L., lesny, J., & Olejnik, J. (2013). Climate change impact on development rates of the codling moth (Cydia pomonella L.) in the Wielkopolska region, Poland. International Journal of Biometeorology, 57, 31–44. Karlsson, B. (2014). Extended season for northern butterflies. International Journal of Biometeorology, 58, 691–701. Kiritani, K. (2013). Different effects of climate change on the population dynamics of insects. Applied Entomology and Zoology, 48, 97–104. Menendez, R., Gonzalez-Megias, A., Collingham, Y., Fox, R., Roy, D. B., Ohlemuller, R., & Thomas, C. D. (2007). Direct and indirect effects of climate and habitat factors on specialist and generalist butterfly diversity. Ecology, 88, 605–611. O’Connor, M. I., Selig, E. R., Pinsky, M. L., & Altermatt, F. (2012). Toward a conceptual synthesis for climate change responses. Global Ecology and Biogeography, 21, 693–703. O’Neill, B. F., Bond, K., Tyner, A., Sheppard, R., Bryant, T., Chapman, J., … Donnelly, A. (2012). Climatic change is advancing the phenology of

moth species in Ireland. Entomologia Experimentalis et Applicata, 143, 74–88. Roy, D. B., & Sparks, T. (2000). Phenology of British butterflies and climate change. Global Change Biology, 6, 407–416. Van Asch, M., & Visser, M. E. (2007). Phenology of forest caterpillars and their host trees: The importance of synchrony. Annual Review of Entomology, 52, 37–55. Van Dyck, H., Bonte, D., Puls, R., Gothard, K., & Maes, D. (2015). The lost generation hypothesis: Could climate change drive ectotherms into a developmental trap? Oikos, 124, 54–61. Visser, M. E., & Holleman, L. J. M. (2001). Warmer springs disrupt the synchrony of oak and winter moth phenology. Proceedings of the Royal Society of London B: Biological Sciences, 268, 289–294. Westgarth-Smith, A. R., Roy, D. B., Scholz, M., Tucker, A., & Sumpter, J. P. (2012). The role of the North Atlantic Oscillation in controlling U.K. butterfly populations size and phenology. Ecological Entomology, 37, 221–232. Williams, D. W., & Liebhold, A. M. (2002). Climate change and the outbreak ranges of two North American bark beetles. Agricultural and Forest Entomology, 4, 87–99.

How to cite this article: Blackshaw RP, Esbjerg P. Statistical indicators for insects caught in the development trap. J Appl Entomol. 2018;142:272–276. https://doi.org/10.1111/ jen.12403