Possible causes of the decline of the butterfly species
Pyronia tithonus in the Netherlands
Bachelor Biology
Institute for Biodiversity and Ecosystem dynamics
Author:
A.J. (Amke Johanna) Jap Tjoen San (11640812)
Supervisors:
Astrid Groot
Chris van Swaay (Senior project manager Dutch butterfly conservation)
Date:
Abstract
Butterflies play a major role in the implementation of fundamental ecological roles. However, unfortunately the majority of the butterfly species in the Netherlands have been declining since 1992 (van Swaay, 2020). Many studies that investigate the explanation for these declines cover specialist butterfly species. These type of species have always showed to be more sensitive to changes and showed heavier declines (Ghazanfar et al., 2016). However, in the last two decades, fifty-five percent of all widespread species have shown significant declines in the Netherlands (van Swaay, 2020; Ghazanfar et al., 2016). The explanation for these declines is unknown for the majority of these species, often because these species are more difficult to model than specialist species (Fernández et al., 2015). Consequently, it is important to start researching the declines of these types of butterfly species.
This research analyses the changing population dynamics of the widespread butterfly species Pyronia tithonus. P. tithonus is among the fifty-five percent of the widespread Dutch butterfly species that have been declining without a known explanation. The species only occurs in northern Netherlands, mainly in the province Drenthe, and in the southern provinces Zeeland, Noord-Brabant and Limburg. The individuals in the northern provinces show steeper declines than the individuals in the southern provinces. This phenomenon does not have a known explanation either. The objective of this research was, therefore, to identify unbeneficial parameters for the population abundance of P. tithonus and to test whether these parameters could explain the heavier decline in the north. Rtrim was used to calculate trends and to test the significance of the parameters.
The results show that the following parameters are unbeneficial to the overall population abundance of P. tithonus: acoustic noise, productivity, nitrate concentration in the groundwater, mean annual precipitation, number of dry days in the year, and ammonia emissions. The following parameters positively affect the total population abundance of P. tithonus: Low values of soil nitrogen from fertilizer, buffered soils, high percentages of low vegetation, and the presence of shrubs.
Moreover, the results suggest that the higher NOx emissions in the northern provinces can be the explanation
for the heavier decline in the north. However, further research into the differences between the individuals living in the northern and southern provinces is recommended, given that NOx emissions did not show to be significant for the
total population abundance. Consequently, this research suggests that the individuals in the north could differ from the individuals in the south in the way they react to their environment. This could be caused by genetic differences between the individuals in the north and in the south. However the results of this analysis alone are not sufficient to draw this conclusion.
Therefore, the results of this research motivates further research into the genetic differences between the northern and southern individuals of P. tithonus. Moreover, it motivates the need for further research into the changing population dynamics of widespread species, because a lot of knowledge on these types of species is still lacking.
Keywords: Pyronia tithonus, The hedge brown, The gatekeeper, Butterflies, population dynamics, population trends, conservation biology, distribution, ecology, parameters, monitoring data, rtrim, habitat.
Introduction
Butterflies are ecologically valuable organisms. They play a major role in the pollination of natural and agricultural vegetation, they form a food source for multiple organisms, they contribute to the evolution of plants, they provide recreational value, and their abundance is an indication of habitat health (Ghazanfar, Malik, Hussain, Iqbal, & Younas, 2016; da Silva Santos, de Oliveira Milfont, Silva, Carneiro & Castro, 2020; Wood & Pullin, 2002; Evans et al., 2020).
Unfortunately, the majority of the butterfly species has been declining in the Netherlands in the last decade (van Swaay, 2020). Studies that analyzed the declines of butterfly species in the past often investigated the population dynamics of specialist butterfly species. These types of butterfly species show to be more sensitive, and therefore experience heavier declines than widespread species. However, this has changed in the last decade for Dutch butterfly species (van Swaay, 2020; Ghazanfar et al., 2016). Fifty-five percent of Dutch widespread species have shown significant declines (van Swaay, 2020; Ghazanfar et al., 2016). The explanation for the declines, especially for widespread species, is often unknown. This is because these species are more difficult to model than specialist species (Fernández, Jordano, & Haeger, 2015). Consequently, it is important to investigate the causation of the declines of these types of butterfly species.
Figure 1. A map of the occurrence of Pyronia tithonus in the Netherlands from 1993 until 2003 (left), 1981 until 1994 (right) and from before 1981 (middle). Source: Bos et al., 2006.
Here, I have investigated the changing population dynamics of Pyronia Tithonus, otherwise known as the Hedge Brown, in the Netherlands. P. tithonus is a beach butterfly that lives in semi-shaded rough grasslands. It can also reside in forests and forest edges, on dunes, near ditches, and near roadsides (Dutch butterfly conservation, n.d.). Their flight period is relatively short and ranges from July 16th up to August 15th (Bos et al., 2006; Dutch butterfly conservation, n.d.). The species produces one generation per year and goes into hibernation in October as a caterpillar (Bos et al., 2006; Dutch butterfly conservation, n.d.). It is a generalist species. This entails that the species does not have any specific preferences for diet. P. tithonus feeds on multiple narrow-leaved grass species in the caterpillar stage (Dutch butterfly conservation, n.d.) and adult butterflies do not have specific preferences towards flowering plants for nectar (Dutch butterfly conservation, n.d.). Its mobility is dependent on high vegetation, the ability for shelter, and nectar availability (Merckx & van Dyck, 2002; Bos et al., 2006; Dutch butterfly conservation, n.d.). It is a mobile species, and males are generally more mobile than females (Bos et al., 2006). The species is present in Europe, Southeast Asia, and Australia, and is most abundant in temperate climates (Global Biodiversity Information Facility, 2020).
P. tithonus is currently categorized as a sensitive species, in the red list of butterfly species in the Netherlands (Dutch butterfly conservation, n.d.; van Swaay, 2019). It is a generalist grassland species, that does not have many known specific requirements for its habitat (Bos, Bosveld, Groenendijk, van Swaay, & Wynhoff, 2006; Dutch butterfly conservation, n.d.). The known necessities are, that the preferred habitat is rough, contains lots of shade, and herbaceous plants. Therefore, this species can reside in a range of habitat types, mentioned before (Dutch butterfly conservation, n.d.).
P. tithonus shows a distinctive distribution in the Netherlands (see figure 1). It only occurs in the Northern Netherlands, mostly in the province Drenthe, and in the southern provinces Zeeland, Noord-Brabant, and Limburg (Bos et al., 2006; Dutch butterfly conservation, n.d.). Such a distribution is not uncommon for P. tithonus. All over Europe, the species is known to be abundant at one location, and absent a few meters further even though the habitat seems similar (Merckx & van Dyck, 2002; Bos et al., 2006).
Figure 2. Population trend of P. tithonus north (orange) and south (blue) of the big rivers. Source: NEM (the Dutch Butterfly Conservation, CBS).
Moreover, this species shows contrasting population trends. The total population trend in the Netherlands indicates a decline. However, the trend shows a sharper decline in the northern provinces (see figure 2) (Bos et al., 2006;Dutch butterfly conservation, n.d.). Little prior research has looked for specific factors that could explain the declining trend of the total population of P. tithonus. Moreover, the explanation for the difference in trends between north and south has not been found. In this thesis, I aim to identify specific parameters that influence the abundance of P. tithonus and whether these parameters can explain the heavier decline in the Northern provinces.
This research was in coorperation with the senior project manager of the Dutch Butterfly Conservation, Chris van Swaay. The Dutch Butterfly Conservation is an organization that combines Dutch and European knowledge of butterflies and dragonflies. This knowledge is used for the conservation and research of species. Their objective is to increase the abundance of butterflies in the Netherlands (Dutch Butterfly Conservation, n.d-a).
The analysis, executed in this research, has incorporated fourteen parameters of which the influence on the total, the northern and the southern trend of P.tithonus was tested: nitrogen deposition, acoustic noise, productivity (defined with captured CO2), maximum annual temperature, annual mean precipitation, number of dry days in the
year, nitrate concentration in the groundwater, nitrogen oxide emissions, ammonia emissions, nitrogen in soil from fertilizer, whether or not the soil is a buffer soil, percentage of shrubs, percentage of low vegetation, and percentage of trees.
First the parameter nitrogen deposition was researched. Species richness is significantly threatened by nitrogen deposition. When atmospheric nitrogen settles on the soil, it causes fertilization and acidification which reduces productivity (World Wide Fund, 2020; Chen, Lan, Bai, Grace, & Bai, 2013). This changes the chemical composition of vegetation, and causes the decline of plant species richness (World Wide Fund, 2020). These changes have impact on animal populations, and especially pollinators show to be sensitive to nitrogen deposition (World Wide Fund, 2020; see appendix III).
Next, the parameter acoustic noise. If butterfly species live near airports or other sound producing infrastructure, they encounter many types of unnatural circumstances, such as noise. Noise could have an effect on how butterfly species behave and could therefore influence their abundances negatively (Peterson, 2019).
Then the influence of the parameter productivity was investigated. In this research, productivity of vegetation is defined as the rate of captured carbon dioxide. This is a criterion for the amount of energy present in the system. More energy, i.e. a high primary production, equals a higher species richness (Bailey et al., 2004). Which is why this parameter will be tested for P. tithonus.
Next, the three weather parameters: maximum temperature, mean precipitation and drought were researched.
Butterflies are poikilothermic organisms (Stefanescu et al., 2003; Beaumont & Hughes, 2002; Fernández et al., 2015). This entails that they are sensitive to changes in climate. Changes in temperature can cause physical and phenological changes (Stefanescu et al., 2003; Beaumont & Hughes, 2002; Fernández et al., 2015; Posledovich et al., 2020). Physical changes being, for example, changes in body size, and phenological changes being for example, first appearances of adult butterflies (Sparks & Yates, 1997; Fisher & Karl, 2010). Moreover, the technical analysis by van Swaay (2019) established that maximum annual temperature is of high importance to P. tithonus, it scored an importance of ninety percent. The effect of temperature on the population of P. tithonus will therefore be tested.
Precipitation is another climatic factor that can affect the phenology of butterfly species and high rates of precipitation can cause increased mortality (Mills et al., 2017; Oberhauser & Peterson, 2003). Moreover, annual mean precipitation was included in the technical report of van Swaay (2019) and scored an importance of a little over fifty percent.
Drought has been proved to have a negative effect on butterfly populations It is said to reduce breeding successes and oviposition (i.e. egg deposition) (Pollard, Greatorex-Davies, & Thomas, 1997; Salgado, DiLeo, & Saastamoinen, 2020). Moreover, drought was included in the technical report of van Swaay (2019) and scored an importance of sixty percent. Therefore, this research will test the influence of drought on the population of P. tithonus. Then, the influence of the parameter nitrates in the groundwater was researched. The nitrate concentration present in groundwater is caused by the washing out of excessive fertilizer (World Wide Fund, 2020). Fertilizer has the same effect on the soil and on vegetation as nitrogen deposition. It changes the composition of the vegetation and causes acidification. Therefore, the concentration of nitrate in groundwater will be tested for P. tithonus.
Next, the influence of two air pollution parameters was researched.
Air pollutants like NOx emissions are known to affect the metabolism and carbon fixation of plants (Weber,
Tingey, & Andersen, 2002). This has detrimental effects on the growth development of plants and changes their composition (Weber, Tingey, & Andersen, 2002). Butterflies are dependent on vegetation, because they have a mutualistic relationship with each other. Therefore, this can also affect butterfly species negatively. Therefore, the effect of this parameter was tested on P. tithonus specifically.
Just like with NOx emissions, ammonia emissions can cause the composition of plants to change, due to the
interference of growth development, which can affect butterflies negatively (Weber et al., 2002). Therefore, the effect of this parameter was tested on P. tithonus specifically.
Then, the parameter nitrogen in the soil, originated from fertilizer, was researched. Nitrogen originated from fertilizer has the same effect on the soil and on vegetation as nitrogen deposition (World Wide Fund, 2020). It can cause acidification and change the vegetation composition. Therefore, the effect of soil nitrogen, originated from fertilizer, will be tested for P. tithonus.
Next, the parameter buffer soils was investigated. There are five major types of soils that present in the Netherlands: sea clay, river clay, calcareous löss, dune and higher sand. Out of the five, higher sand soils are the only soil type that is categorized as sensitive to acidification given their deficiency of buffering capacity, for example due to low quantities of calcareous compounds (de Vries, Breeuwsma, & de Vries, 1989). Whether the soil is buffered or not, determines if the soil can undergo nitrogen deposition without losing its plant and animal species richness.
And lastly, the influence of three vegetation parameters was investigated.
In the study of Van Dyck et al. (2009) the recent declines of fifty five percent of the widespread species in the Netherlands was studied. They concluded that the highest declines occurred in farmland, urban, and particularly woodland areas. Besides this, multiple studies have concluded that high coverage of trees affects butterfly species negatively (Öckinger, Eriksson, and Smith, 2006; Öckinger, Hammarstedt, Nilsson, & Smith, 2006-a). However, for some species it has positive effects (Öckinger et al., 2006).. Therefore, the effect was tested on P. tithonus specifically. For P. tithonus specifically, shrubs and high vegetation are necessary for the ability for micro-distribution (Bos et al., 2006; Merckx & van Dyck, 2002; Dutch Butterfly Conservation, n.d.). P. tithonus use shrubs and high vegetation as shelter, for example when males are searching for mates over relatively long distances (Bos et al., 2006; Dutch Butterfly Conservation, n.d.; Merckx & van Dyck, 2002). However, low vegetation can also be preferred by this species, given that low vegetation are often herbaceous plant or grass species. These types of plant species are often beneficial in the habitat of P.tithonus, because most grass species can be host plants for this species and herbaceous plants often co-occur with flowering plants (Dutch Butterfly Conservation, n.d.; Salgado et al., 2020). Therefore, the effect of the percentage of shrubs and the percentage of low vegetation was tested on the population abundance of P.tithonus in the Netherlands.
The results of this research can give insights into specific habitual factors that are unbeneficial to the abundance of P.tithonus, for the specific areas in which the species occurs. It can motivate conservation strategies for the improvement of the habitat for this and possibly other species and it can motivate further research concerning the population dynamics of other widespread butterfly species.
Methods
Monitoring data of P. tithonus
To assess population trends, the monitoring data of P. tithonus, which include the species count, the site numbers, and the date of the observation, was obtained from The Dutch Butterfly Conservation.
This dataset contains many missing values, which is a frequently occurring phenomenon with organism monitoring data (Pannekoek, Bogaart, & van der Loo, 2018). This type of data is dependable on volunteers to report their observations and this does not happen regularly. To be able calculate trends and indices for this data, the package rtrim was used in the statistical program R version 1.2.5042 (Pannekoek et al., 2018). Rtrim uses several log-linear models to predict the missing values, by estimating the observed counts (Pannekoek et al., 2018). Moreover, it is possible to assign covariables to the rtrim model, to test whether specific parameters have a significant influence on the population abundance and to see whether it causes an increase or a decrease of the trend (Pannekoek et al., 2018).
To assess what the population dynamics look like and to apply the analysis on all the areas in which P.tithonus occurs, the total population trend, the southern and the northern trend were calculated and plotted using rtrim. Areas with the lowest occurrence of P. tithonus and determining parameters
To assess which parameters could have a negative influence on the population of P. tithonus, the areas of lowest occurrence were determined. And to determine these areas, I used a species distribution model (SDM). This SDM was provided by Chris van Swaay, senior project manager at the Dutch Butterfly Conservation. The SDM data was modified with the statistical program R version 1.2.5042, to display the areas in which the chance of occurrence is less than fifty percent (see figure 3). This percentage was chosen because no prior research mentioned certain percentages of P. tithonus that are considered as a low occurrence. Consequently, the most general partition was chosen, a partition at the mean of fifty percent.
Figure 3. Map of the Netherlands with blue points that indicate the sites in which P.tithonus has a lower chance of occurrence than fifty percent.
To determine possible parameters that could affect the abundance of P. tithonus negatively, the sites with the lowest occurrence of p. tithonus have been categorized by land-cover, using the Corine Land Cover map of 2018 obtained from Copernicus Land Monitoring Service. Copernicus is a European program that monitors the planet. To provide reliable and up to date data, they combine satellite- and monitoring data to create maps regarding the following sectors: land, marine, atmosphere, climate change, emergency management, and security.
To define which parameters could cause the low occurrence in these sites and possibly cause the decline of the population abundance in the Netherlands in total, literary research was performed. Table 7 in appendix I, shows the twenty-six parameters that are correlated with these land-cover types. Usable data could only be obtained on fourteen of these parameters, and these were used in this analysis: nitrogen deposition, acoustic noise, productivity (defined with captured CO2), maximum annual temperature, annual mean precipitation, number of dry days in the
year, nitrate concentration in the groundwater, nitrogen oxide emissions, ammonia emissions, nitrogen in soil from fertilizer, whether or not the soil is a buffer soil, percentage of shrubs, percentage of low vegetation, and percentage of trees.
Population trends with parameters
To test if the abovementioned parameters have influence on the total population of p. tithonus, all the parameters that were obtained from the RIVM, KNMI, Atlas Living environment, and the Dutch butterfly conservation databases, have been tested with the total population trend.
Moreover, to determine whether these parameters can explain the heavier decline in the northern provinces, all the parameters have been tested for the individuals living in the northern and the southern provinces. To add the parameters to the rtrim model, the data needs to be binary (Pannekoek et al., 2018). This has been done by assigning a “high” value to numbers above the median, and a “low” value to numbers below the median. The significance of the parameters was tested with the Wald test. This test determines the effect of a parameter on each of the trends, by calculating if the slopes of the “high” values significantly differ from the slopes of the “low” values (Pannekoek et al., 2018).
The parameters
To test whether they affect the population abundance of P. tithonus, the following fourteen parameters were used as covariables: nitrogen deposition, acoustic noise, productivity (defined with captured CO2) , maximum annual
temperature, annual mean precipitation, number of dry days in the year, nitrate concentration in the groundwater, nitrogen oxide emissions, ammonia emissions, nitrogen in soil from fertilizer, whether or not the soil is a buffer soil, percentage of shrubs, percentage of low vegetation, and percentage of trees. Table 7 in appendix I shows the literary articles from which parameters were drawn up. Not all parameters were implemented in this analysis, because data could not be obtained for some of them.
1. Nitrogen deposition
Since nitrogen has been found to be an important parameter that affects butterflies negatively (World Wide Fund, 2020). I obtained the map of nitrogen deposition in the Netherlands from the RIVM database that includes the NH3
concentration and the wet deposition of NHx in the year 2018 (Hoogerbrugge et al., 2019). This dataset was chosen
because it was the most recent data that could be obtained, and that was usable for this analysis.
To be certain that the dataset was split in two equally large datasets, which would make the analysis more reliable, I first plotted the distribution of the parameter nitrogen deposition (see Figure 4). Then I determined the median value of 1681 mol/ha/yr and converted the parameter data into binary data with two categories, a “low” category with values < 1681 and a “high” category with values of 1681 mol/ha/yr or more.
Figure 4. Histogram of the covariable nitrogen deposition, the unit is in mol/ha.yr (Hoogerbrugge et al., 2019). 2. Acoustic noise
To test whether acoustic noise, the type of noise that can be heard, has an influence on P. tithonus living near airports, but also near roads and other noise-producing infrastructure, I used a map of the noise from all sources in the Netherlands (Peterson, 2019). I obtained this map from the Atlas living environment database, which includes the noise in dB produced from roads (measured in 2017), railways (measured in 2016), aviation (measured in 2011), industry (n.d.) and wind turbines (measured in 2015) (Atlas Living Environment, n.d. ; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To divide the dataset in two equally large datasets, which would make the analysis more reliable, I first plotted the distribution of the noise parameter (see figure 5). Then I determined the median value of 51.50 dB and converted the parameter data into binary data with two categories, a “low” category with vales < 51.50 dB and a “high” category with values of 51.50 dB or more.
Figure 5. Histogram of the covariable acoustic noise, the unit is in dB (Atlas Living Environment, n.d.). 3. Productivity
To test whether productivity has and influence on P. tithonus, the productivity, measured by the captured CO2, is
implemented in this analysis. The map for this data was obtained from the RIVM database, and contains the amount of captured carbon by vegetation in the Netherlands in the year 2013 (RIVM, 2013; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To be certain that the dataset would be split in two datasets of equal size, I first plotted the distribution of the productivity parameter (see figure 6), after which I determined the median value of 3.8 C/m2/year. Then I
converted the parameter data into binary data with two categories, a “low” category with values < 3.8 C/m2/year and
a “high” category with values of 3.8 C/m2/year or more.
Figure 6. Histogram of the covariable productivity (measured with CO2 capture), the unit is in ton C/m2/yr (RIVM, 2013). 4. Maximum annual temperature
Since temperature is known to affect most butterfly species, the influence of this parameter was tested on P.tithonus (Posledovich et al., 2020). The map of the maximum annual temperature was obtained from the KNMI database, and contains information of the maximum annual temperature calculated for the period of 1981 until 2010 (KNMI, 2012; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To divide the dataset in two equally large datasets, I first plotted the distribution of the temperature parameter (see figure 7). Then I calculated the median value to be 0.008681 (0.001)˚C, which equals 8.6 ˚C, and converted the parameter data into binary data with two categories, a “low” category with values < 0.008681 ˚C and a “high” category with values of 0.008681 ˚C or more.
Figure 7. Histogram of the covariable maximum annual temperature, the unit is in 0.001 degrees (KNMI, 2012). 5. Annual mean precipitation
Just like temperature, precipitation is another weather related parameter that has considerable influence on butterfly species (Mills et al., 2017). Therefore, this was used as a parameter for the analysis of P.tithonus. The map of the annual mean precipitation was obtained from the KNMI database and contains information on the mean precipitation in the Netherlands, calculated for the period of 1981 until 2010 (KNMI, 2017; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To divide the dataset in two equally large datasets, I first plotted the distribution of the precipitation parameter (see figure 8). Then I calculated the median value to be 306.3 mm, after which I converted the parameter data into binary data with two categories, a “low” category with values < 306.3 mm and a “high” category with values of 306.3 mm or more.
Figure 8. Histogram of the covariable mean annual precipitation, the unit is in mm (KNMI, 2017). 6. Drought
The last weather related parameter used in this analysis is drought. The drought map was obtained from the KNMI database, and includes the mean number of dry days in a year, calculated over the period 1981 until 2010 (KNMI, 2017; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To divide the dataset in two equally large datasets, I first plotted the distribution of the drought parameter (see figure 9). Then I calculated the median value to be 306.3 mm, after which I converted the parameter data into binary data with two categories, a “low” category with values < 306.3 mm and a “high” category with values of 306.3 mm or more.
Figure 9. Histogram of the covariable number of dry days (KNMI, 2017). 7. Nitrate concentration in groundwater
To test the effect of nitrate in the groundwater on P.tithonus, this parameter was used in this analysis. The map for this parameter was obtained from the RIVM database, and includes the information of the nitrate concentration in the groundwater in the Netherlands in the period 2012 up to 2015 (RIVM, n.d.; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To divide the dataset in two equally large datasets, I first plotted the distribution of this parameter (see figure 10). Then I determined the median value of 21.884 mg/L and converted the parameter data into binary data with two categories, a “low” category with values < 21.884 mg/L and a “high” category with values of 21.884 mg/L or more.
Figure 10. Histogram of the covariable nitrate concentration in the groundwater, the unit is in mg/L (RIVM, n.d.). 8. Nitrogen oxide (NOx) emissions
Since nitrogen oxide emission is known to affect species richness (World Wide Fund, 2020; Kopáček & Posch, 2011). This was used as a parameter in this analysis. The nitrogen map was obtained from the RIVM database, and contains the information of the amount of nitrogen per square kilometer for the year 2017 (RIVM, n.d.-a ; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To divide the dataset in two equally large datasets, I first plotted the distribution of the NOx emission
parameter (see figure 11). Then I determined the median value of 518987 kg/km2 and converted the parameter data
into binary data with two categories, a “low” category with values < 518987 kg/km2 and a “high” category with values
Figure 11. Histogram of the covariable NOx emissions, the unit is in kg/km2 (RIVM, n.d.-a). 9. Ammonia emissions
Just like NOx emissions, ammonia emissions have considerable effect on vegetation and therefore affect butterfly
species negatively (World Wide Fund, 2020; Kopáček & Posch, 2011). The ammonia map was obtained from the RIVM database and contains the total amount of emitted ammonia in the year 2017 (RIVM, n.d-b.; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To divide the dataset in two equally large datasets, I first plotted the distribution of the ammonia emission parameter (see figure 12). Then I determined the median value of 445795 kg/km2 and converted the parameter data
into binary data with two categories, a “low” category with values < 445795 kg/km2 and a “high” category with values
of 445795kg/km2 or more.
Figure 12. Histogram of the covariable ammonia emissions, the unit is in kg/km2 (RIVM, n.d.-b).
10. Soil nitrogen from fertilizer
To test the effect of fertilizer on P.tithonus, this was used as a parameter in this analysis. The map about soil nitrogen, originating from fertilizer use, was obtained from the RIVM database. It contains information about the total amount of nitrogen in the soil that originated from fertilizer in the Netherlands the year 2017 (RIVM, n.d-c.; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To be certain that the dataset was split in two equally large datasets, I first plotted the distribution of the nitrogen from fertilizer parameter (see figure 13). Then I determined the median value of 233760 kg/km2 and
converted the parameter data into binary data with two categories, a “low” category with values < 233760 kg/km2
Figure 13. Histogram of the covariable soil nitrogen originating from fertilizer, the unit is in kg/km2 (RIVM, n.d.-c). 11. Unbuffered soil
The effect of soil type, mainly whether the soil Is buffered or not, on P. tithonus was investigated. The buffered soils data was obtained from the monitoring data of P.tithonus, provided by the Dutch butterfly conservation. It contains information on the locations of the five major soil types in 2019, which has been categorized as buffered or unbuffered soils (van Swaay, personal communication, June 3, 2020; see appendix III).
The dataset was split into the northern and southern provinces and could therefore not result in two equally large datasets of the soil type. This dataset contains 5264 sites with buffered soils and 11536 sites with unbuffered soils.
12. Percentage of trees
Since the percentage of trees affects many butterfly species, this parameter was implicated in this analysis (Öckinger et al., 2006; Öckinger et al., 2006-a; van Dyck et al., 2009). The map about the percentage of trees was obtained from the atlas living environment database. It contains information on the percentage of trees in the Netherlands per 10 X 10 meter grid cell in 2017 (Atlas Living Environment, n.d.-a; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To be certain that the dataset was split in two equally large datasets, I first plotted the distribution of the trees parameter (see figure 14). Then I determined the median value of 47.5 percent and converted the parameter data into binary data with two categories, a “low” category with values < 47.5 percent and a “high” category with values of 47.5 percent or more.
13. Percentage of shrubs
Since a high percentage of shrubs is of considerable importance for the micro-distribution of P.tithonus, the specific effect on the population abundance has been tested in this analysis (Bos et al., 2006; Merckx & van Dyck, 2002; Dutch Butterfly Conservation, n.d.). The map about the percentage of shrubs was obtained from the Atlas living environment database. It contains information on the percentage of shrubs in the Netherlands per 10 X 10 meter grid cell in 2017 (Atlas Living Environment, n.d.-b; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To be certain that the dataset was split in two equally large datasets, I first plotted the distribution of the shrubs parameter (see figure 15). Then I determined the median value of 12 percent and converted the parameter data into binary data with two categories, a “low” category with values < 12 percent and a “high” category with values of 12 percent or more.
Figure 15. Histogram of the covariable percentage of shrubs (Atlas Living Environment, n.d.-b). 14. Percentage of low vegetation (excl. arable land)
Just like the shrubs, high vegetation is preferred for the micro-distribution of P.tithonus (Bos et al., 2006; Merckx & van Dyck, 2002; Dutch Butterfly Conservation, n.d.). However, low vegetation can also be preferred by this species, given that low vegetation are often herbaceous plant or grass species. These types of plant species are often beneficial in the habitat of P.tithonus (Dutch Butterfly Conservation, n.d.). Therefore, low vegetation was used as a parameter in this analysis. The map about the percentage of low vegetation was obtained from the Atlas living environment database. It
contains information on the percentage of high vegetation outside of agricultural vegetation in the Netherlands per 10 x 10 meter grid cell in 2017 (Atlas living environment, n.d.-c; see appendix III). This dataset was chosen because it was the most recent data that could be obtained, and that was usable for this analysis.
To be certain that the dataset was split in two equally large datasets, I first plotted the distribution of the low vegetation parameter (see figure 16). Then I determined the median value of 49 percent and converted the parameter data into binary data with two categories, a “low” category with values < 49 percent and a “high” category with values of 49 percent or more.
Results
Population trend, northern trend, and southern trend without covariables
The total population trend of P. tithonus is represented in Figure 17. The population shows an oscillating trend, however, predominantly a declining trend is visible. The y-axis shows indexes. An index is a comparison of the population state of a certain year, with the population state of the previous year, a value of one represents one hundred percent of the abundance (Pannekoek et al., 2018). The first value in the graph is always 100 percent, because this is the first year that observations were made and no comparison can be made with a previous year. This does not mean that this is the maximum value of abundance that a certain species has had in the Netherlands, but it is the first information on the abundance of a species that has been measured.
In total the abundance of P. tithonus declined from one hundred percent in 1992 to sixty-three percent in 2019.
Figure 18 and 19 show the trends of the individuals living in the northern and the southern provinces. These graphs show the steeper decrease in the northern provinces compared to the southern provinces.
Figure 17. Population trend of P. tithonus in the total of the Netherlands. The x-axis shows the years in which the observations were made and the y-axis shows the indexes. Indexes show the state of the butterfly population in comparison with the previous year.
Figure 18. Trend of the individuals of P. tithonus living in the southern provinces. The x-axis shows the years in which the observations were made and the y-axis shows the indexes. Indexes show the state of the butterfly population in comparison with the previous year.
Figure 19. Trend of the individuals of P. tithonus living in the northern provinces. The x-axis shows the years in which the observations were made and the y-axis shows the indexes. Indexes show the state of the butterfly population in comparison with the previous year. Trim test overall population
Out of all fourteen parameters, ten showed a significant effect on the population abundance of P.tithonus: Acoustic noise, productivity, mean annual precipitation, number of dry days, nitrate in the groundwater, nitrogen from fertilizer, unbuffered soils, ammonia emissions, percentage of trees, and percentage of shrubs (see table 1).
Table 2 shows the changes that a high or a low value of a parameter induces on the total population abundance in percentages (see appendix II for the full graphs that show the trendlines). For the parameter buffered soils no high or low category could be assigned to the data, only buffered or unbuffered soils. The change that buffered soils cause are, therefore, included in the ‘Change in trendline of ‘high’ values’ column and the change that unbuffered soils cause in the ‘Change in trendline of ‘low’ values’ column.
Many parameters (acoustic noise, productivity, precipitation, dry days, nitrate concentration in the groundwater, and ammonia emissions) showed a negative effect on the total population abundance. Both ‘high’ and ‘low’ values of these parameters result in a decrease of the total population abundance.
The parameters soil nitrogen from fertilizer, buffered soils, percentage low vegetation, and percentage of shrubs, showed a positive effect on the total population abundance. Low values of soil nitrogen from fertilizer, low percentages of shrubs , buffered soils, and high percentages of low vegetation result in an increase of the total population trend.
Table 1 An overview of the results of the Wald test, performed with all the parameters in combination with the total population trend.
Parameter tested with the Wald test p-value
Nitrogen deposition 0.800521 Acoustic noise 0.0001106985 Productivity 0.01328981 Maximum annual temperature 0.1926204 Mean annual precipitation 0.02001158 Number of dry days in the year 2.627374e-10 Nitrate concentration in the groundwater 1.086105e-05
NOx emissions 0.2616383
Soil nitrogen from fertilizer 6.256951e-11 Buffered or unbuffered soils 1.014329e-09 Ammonia emissions 9.658017e-08 Percentage trees 0.07513039 Percentage shrubs 0.005369535 Percentage low vegetation (excl. arable vegetation) 1.368354e-05
Table 2 An overview of the changes that each parameter induces on the total population abundance of P. tithonus in percentages.
Parameter Change in trendline of ‘high’ values Change in trendline of ‘low’ values
Acoustic noise - 50 % - 20 %
Productivity - 38 % - 60 %
Mean annual precipitation - 49 % - 26 %
Number of dry days in the year - 64 % - 0 %
Nitrate concentration in the groundwater
- 69 % - 31 %
Soil nitrogen from fertilizer - 63 % + 20 %
Buffered or unbuffered soils + 103 % - 52 %
Ammonia emissions - 65 % - 9 %
Percentage low vegetation + 221 % - 34 %
Percentage shrubs + 6 % + 107 %
Trim tests for North and South
Nine parameters showed a significant effect on the abundance of the individuals of P.tithonus, living in the northern provinces: Productivity, maximum annual temperature, annual precipitation, nitrate in the groundwater, buffer soils, NOx emissions, percentage trees, percentage shrubs, and percentage low vegetation (see table 3).
Table 4 shows the changes that a high or a low value of a parameter induces on the abu17ndance of the northern individuals in percentages (see appendix II for the full graphs that show the trendlines). The majority of the parameters, showed a negative effect on the abundance of the northern individuals: productivity, maximum annual temperature, annual precipitation, nitrate in the groundwater, and NOx emissions, percentage trees. Both ‘high’ and
‘low’ values of these parameters result in a decrease of the trend.
The parameters buffered soils and percentage of low vegetation showed a positive effect on the abundance of the northern individuals. Buffered soils and high percentages of low vegetation result in an increase of the trend.
When performing the Wald test on the parameters implemented with the southern dataset, thirteen parameters showed a significant effect on the abundance of the individuals of P.tithonus living in the southern provinces: nitrogen deposition, acoustic noise, maximum annual temperature, annual precipitation, number of dry days, nitrate in the groundwater, buffer soils, NOx emissions, nitrogen from fertilizer, ammonia emissions, percentage
of shrubs and percentage of low vegetation (see table 3).
Table 5 shows the changes that a high or a low value of a parameter induces on the abundance of the southern individuals in percentages (see appendix II for the full graphs that show the trendlines). The minority of the parameters showed a negative effect on the abundance of the northern individuals: nitrogen deposition, acoustic noise, annual precipitation, nitrate concentration in the groundwater, and ammonia emissions. Both ‘high’ and ‘low’ values of these parameters result in a decrease of the trend.
The parameters maximum annual temperature, dry days, NOx emissions, nitrogen from fertilizer, buffer soils,
percentage of shrubs, percentage of low vegetation , and percentage of trees showed a positive effect on the abundance of the southern individuals. Low values of maximum annual temperature, dry days, NOx emissions, soil
nitrogen from fertilizer, buffered soils, low percentages of shrubs and trees, and high percentages of low vegetation result in an increase of the trend.
Table 3 An overview of the results of the Wald test, performed with all the parameters in combination with the northern and the southern trend.
Parameter tested with the Wald test: P-value of northern dataset p-value of southern dataset
Nitrogen deposition 0.2487881 0.015851 Acoustic noise 0.9341708 9.14416e-05 Productivity 0.01316547 0.1618957 Maximum annual temperature 0.000 1.712468e-05 Mean annual precipitation 1.892013e-09 0.009474969 Number of dry days in the year 0.3356248 3.304217e-06 Nitrate concentration in groundwater 5.551115e-16 0.06547187
Soil nitrogen from fertilizer 0.6353142 1.294749e-05 Buffer soils 0.0002303775 2.749728e-06 Ammonia emissions 0.3056459 8.74556e-05 Percentage trees 0.05650997 0.01805295 Percentage shrubs 0.1135157 0.0001937738 Percentage low vegetation
(excl. arable vegetation) 0.0008317921 0.002965177
Table 4 An overview of the changes that each parameter induces on the total population abundance of the individuals of P. tithonus in the northern provinces in percentages.
Parameter in the northern dataset Change in trendline “high” values Change in trendline “low” values
Productivity - 24 % - 72 %
Maximum annual temperature - 39 % - 96 %
Mean annual precipitation - 60 % - 96 %
Nitrate concentration in the groundwater
- 94 % - 12 %
NOx emissions - 52 % - 95 %
Buffered or unbuffered soils + 51 % - 20 %
Percentage low vegetation (excl. arable vegetation)
+ 72 % - 17 %
Table 5 An overview of the changes that each parameter induces on the total population abundance of the individuals of P. tithonus in the southern provinces in percentages.
Parameter in the southern dataset Change in trendline “high” values Change in trendline “low” values
Nitrogen deposition - 71 % - 10 %
Acoustic noise - 71 % - 10 %
Maximum annual temperature - 40 % + 30 %
Mean annual precipitation - 41 % - 3 %
Number of dry days in the year - 55 % + 12 %
Nitrate concentration in the groundwater
- 54 % - 32 %
NOx emissions - 37 % + 2 %
Soil nitrogen from fertilizer - 51 % + 20 %
Buffered or unbuffered soils + 210 % - 62 %
Ammonia emissions - 58 % - 7 %
Percentage shrubs - 10 % + 439 %
Percentage low vegetation (excl. arable vegetation)
+ 231 % - 17 %
Percentage trees - 40 % + 143 %
Differences in parameters between north and south
The majority of the parameters caused the same kind of effect for both the individuals living in the southern and the northern provinces. This entails that if a parameter causes a decline of abundance in the northern provinces it also causes a decline in the southern provinces. The only parameter that showed contrasting effects is the parameter maximum annual temperatures. Lower maximum annual temperatures cause a decrease in the northern provinces and an increase in the southern provinces according to the results of this analysis.
Not many differences between the southern and the northern dataset can be discovered in this manner. However, when looking at the difference between the significant parameters in the northern and southern provinces, seven parameters show observable differences between the north and the south: Nitrogen deposition, maximum annual temperature, mean annual precipitation, NOx emissions, Nitrogen from fertilizer, buffered soils and ammonia
emissions (see table 6).
Maximum annual temperatures, mean annual precipitation, NOx emissions, nitrogen from fertilizer, and
temperature, mean annual precipitation, and NOx emissions have a significant influence on the abundance of the
individuals living in the northern provinces (see table 3).
Nitrogen deposition and the percentage of buffered soils are higher in the southern provinces. And both of these parameters have significant influence on the abundance of the individuals living in the southern provinces. Table 6 An overview of the different values of the parameters in the northern and the southern provinces.
Parameter Northern provinces Southern provinces
Nitrogen deposition (mol/ha/yr) 1629.685 1777.788
Noise (dB) 52.53846 53.69869
Productivity (ton C/m2 /yr) 4.153762 3.855074
Maximum annual temperature (0.001
˚C) 0.02968469 0.0257843
Mean annual precipitation (mm) 338.5631 324.5716
Number of dry days (nr of days) 46.85119 45.95655
Nitrate concentration in groundwater
(mg/L) 37.65072 35.73704
NOx emissions (kg/km2) 758677 689670.9
Nitrogen from fertilizer (kg/km2) 509323.3 224151.1
Percentage buffered soils (%) 5 36.5
Ammonia emissions (kg/km2) 834113.9 446263
Percentage shrubs (%) 13.67568 15.58197
Percentage low vegetation (%) 50.33898 49.75664
Discussion
By adding all the parameters to the rtrim model, I found that the following parameters have a negative influence on the total abundance of P. tithonus: acoustic noise, productivity, precipitation, dry days, nitrate concentration in the groundwater, and ammonia emissions.
adding the parameters the distribution of P.t. between the Northern and the Southern provinces of the Netherlands, I found that (summary of your main findings, which you then further discuss below)
Unbeneficial Parameters
The first part of the research question was: “Which parameters are unfavorable for the population of P. tithonus in the Netherlands?”. By analyzing fourteen parameters, ten showed to be of significant importance to the total population abundance: acoustic noise, productivity, nitrate concentration in groundwater, mean annual precipitation, number of dry days in the year, ammonia emissions, soil nitrogen from fertilizer, buffered soils, percentage of trees, and percentage of shrubs. Many of these parameters (acoustic noise, productivity, precipitation, dry days, nitrate concentration in the groundwater, and ammonia emissions) showed a negative correlation with the total population abundance.
Acoustic noise is the first parameter that showed to be significant to the total population abundance of P. tithonus. The results of this research suggest that both high and low levels of acoustic noise are unbeneficial to the total abundance of P. tithonus. If there is a lot of noise in the landscape, it affects butterfly species negatively (Bailey et al., 2004). For butterflies a low amount of noise could still be loud in their natural landscape, because the noise in this analysis is originated from roads, aviation and industry, which are all sources that produce loud noise comparing to natural habitats. Therefore, both high and low levels of noise could cause a decrease of the population abundance of P. tithonus. However, these results should be taken into careful consideration when drawing a conclusion, because acoustic noise can also be affected by other factors that could have a contribution to the decreasing population abundance of P. tithonus. Examples of other factors that can affect the abundance of P. tithonus, could be the emission from the cars or factories, waste deposition by the industries, or fragmentation caused by the construction of roads, airports or factories.
The second parameter that showed to be significant to the total population abundance of P. tithonus, is productivity. The results suggest that both high and low values of productivity have a negative effect on the abundance. The unbeneficial effect of a high productivity seems peculiar at first as a high level of productivity often equals a high species richness (Peterson, 2019). However, for this specific species, this could be explained by the fact that this species benefits from shade (Dutch Butterfly Conservation, n.d.; Bos et al., 2006). If vegetation is located in the shade, the productivity (measured in this research with CO2 capture) will be lower, given that light is acquired for
primary production. This could, therefore, explain the unbeneficial effect of high levels of productivity on P. tithonus. The results also suggest that a low value of productivity is unbeneficial. Low values of productivity are often correlated with a low species richness (Bailey et al., 2004). Therefore, both high and low levels of productivity can cause a decrease of the population abundance of P. tithonus. However, these results should be taken into careful consideration when drawing a conclusion, because productivity can also be affected by other factors that could have a contribution to the decreasing population abundance of P. tithonus. One example of such a factor is the amount nutrients in the soil. If there are more nutrients present in the soil productivity is higher, given that nutrients are used by plants for primary production.
The third parameter that showed to be significant to the total population abundance of P. tithonus, is nitrate concentration in the groundwater. The results suggest that both high and low concentrations of nitrate in the groundwater, are negatively correlated with the abundance of P. tithonus. Nitrate ends up in the groundwater through the washing out of excessive fertilizer (World Wide Fund, 2020). Fertilizer has an acidifying effect on the soil, which affects vegetation (World Wide Fund, 2020). Plant communities and compositions could change due to the acidified soils. Butterflies have a mutualistic relationship with plants. As mentioned before, butterflies are important to plants because they aid in their reproduction by pollinating them (Ghazanfar et al., 2016).). Plants help butterflies, because they are a source of food and shelter (Bos et al., 2006; Dutch Butterfly Conservation, n.d.). Moreover, plants and butterflies co-evolve together (Ghazanfar et al., 2016). The negative effect of excessive use of fertilizer on plants could, therefore, explain the unbeneficial effect of this parameter to the abundance of P. tithonus. However, these results should be taken into careful consideration when drawing a conclusion, because nitrate can also be end up in the groundwater from different sources. A few of the possible sources are sewage effluent, cattle and poultry manure (Widory et al., 2004). If these sources are the actual cause of the nitrate in the groundwater, fertilizer alone could have different effects on the population abundance of P. Tithonus.
The fourth parameter that showed to be significant to the total population abundance of P. tithonus, is mean annual precipitation. The results suggest that precipitation unbeneficial for the abundance of P. tithonus. Precipitation is a climatic factor that affects the phenology of butterfly species (Mills et al., 2017). Moreover, excessive precipitation can cause increased mortality of butterfly populations (Oberhauser & Peterson, 2003). This could, therefore, explain
the negative effect of precipitation on the abundance of P. tithonus in the Netherlands. However, these results should be taken into careful consideration when drawing a conclusion, because mean precipitation covers the most basic information on precipitation in the Netherlands. Other precipitation parameters that could contribute to the decrease of the abundance of P. tithonus are annual maximum precipitation, annual minimum precipitation, and a parameter that covers information on the precipitation per month.
The fifth parameter that showed to be significant to the total population abundance of P. tithonus, is the number of dry days per year. The results suggest that the population abundance of P. tithonus is negatively affected by a high number of dry days in the year. Drought is said to reduce breeding successes and oviposition (Pollard et al., 1997; Salgado et al., 2020). Therefore, a longer period of drought could affect the abundance of this species negatively. However, these results should be taken into careful consideration when drawing a conclusion, because other factors also contribute to the sensitivity of and ecosystem to drought. An example of such a factor is the ability of the vegetation in the habitat to withstand drought. If the habitat contains a lot of drought-resistant vegetation, butterfly species will also be less sensitive to drought.
The sixth parameter that showed to be significant to the total population abundance of P. tithonus, is ammonia emissions. The results suggest that both high and low values of ammonia emissions are unbeneficial for the abundance of P. tithonus. Emissions often cause changing plant compositions, due to the interference of growth development and NOx emissions specifically can cause biodiversity loss (Kopáček & Posch, 2011; Weber et al., 2002).
This affects butterfly species negatively, which could explain why both high and low amounts of ammonia emissions are unbeneficial to the population abundance of P. tithonus in this analysis. However, these results should be taken into careful consideration when drawing a conclusion, because there are many other factors that could influence negative effect of ammonia to the abundance of P. tithonus. An example of such a factor is precipitation. Precipitation takes up ammonia from the atmosphere and deposits it on the soil (World Wide Fund, 2020). Areas where ammonia emissions are higher, could also have a higher amount of ammonia deposition. This affects vegetation and could, therefore, also affect the abundance of P. tithonus.
The seventh parameter that showed to be significant to the total population abundance of P. tithonus, is soil nitrogen from fertilizer. The results suggest that the abundance of P. tithonus shows a decline with a high degree of fertilizer use. Fertilizer has a beneficial effect on vegetation, because its nutrients can support plant growth. However, excessive use causes acidification of the soil (World Wide Fund, 2020). This affects vegetation and can cause changes in plant composition. Therefore, excessive use of fertilizer can affect butterfly species negatively. This could explain why high degrees of soil nitrogen, originated from fertilizer, is unbeneficial for the abundance P. tithonus. However, these results should be taken into careful consideration when drawing a conclusion, because other factors could influence the acidification of the soil due to fertilizer. One of these factors is whether the soil contains components that have buffering capacity. This parameter was also implemented in the rtrim model and will be discussed next.
The eighth parameter that showed to be significant to the total population abundance of P. tithonus, is buffered soils. Buffered soils were positively correlated with the abundance of P. tithonus in this analysis. This can be explained by the resistance of these types of soils to acidification. Many soils in the Netherlands have undergone acidification, due to for example nitrogen deposition or excessive fertilizer use (World Wide Fund, 2020). Acidification causes the modification of plant compositions and therefore affects butterfly species. Buffered soils can withstand acidification without loss of productivity, because they have buffering capacity. This buffering capacity could for example be present if the soil contains calcareous components. Having a buffered soil could, therefore, explain why buffered soils are beneficial and unbuffered soils are unbeneficial for the population abundance of P. tithonus. However, these results should be taken into careful consideration when drawing a conclusion, because some soils that are categorized as unbuffered soils (i.e. sandy soils) could still contain some buffering capacity, which could compensate for the negative effects acidification can have on the soil. Therefore, another factor could have contributed to the decline of the abundance of P. tithonus.
The eighth parameter that showed to be significant to the total population abundance of P. tithonus, is the percentage of shrubs. Both high and low percentages of shrubs were positively correlated with the population abundance of P. tithonus in this analysis. P. tithonus needs shrubs, during micro-distribution. For example when males are searching for mates while moving through their habitat, they use shrubs as shelter (Bos et al., 2006; Dutch Butterfly Conservation, n.d.). Therefore, micro-distribution is important for finding mates, and thus for the growth of the population (Dutch Butterfly Conservation, n.d.). Which could explain the beneficial effect of the presence of shrubs for the abundance of P. tithonus. However, these results should be taken into careful consideration when drawing a conclusion, because other types of vegetation can also be used as shelter during micro-distribution. Therefore, sites with low percentages of shrubs could also contain some of these other types of vegetation that could form an opportunity for shelter. This could, consequently, contribute to the increase of the abundance of P. tithonus in these sites.
And the Last parameter that showed to be significant to the total population abundance of P. tithonus, is the percentage of low vegetation. The results show that high percentages of low vegetation were positively correlated with the population abundance of P. tithonus. In this analysis low vegetation entails all vegetation below one meter, excluding agricultural vegetation. The majority of the vegetation with this size, are herbaceous plant or grass species.
The majority of the grass species can be used as host plants for P. tithonus, given that it is a generalist species (Dutch Butterfly Conservation, n.d.). Moreover, herbaceous plants are found to be of importance for the presence of P. tithonus and often co-occur with flowering plants (Dutch Butterfly Conservation, n.d.; Kithara, Yumoto, & Kobayashi, 2008; Salgado et al., 2020). This could, therefore, explain the positive effect of high percentages of low vegetation on the abundance of P. tithonus. However, these results should be taken into careful consideration when drawing a conclusion, because high percentages of low vegetation could also be unbeneficial to P. tithonus, given that this species needs high vegetation for shelter during micro-distribution.
The explanation for the difference in trends
The second part of the research question was: “Can these parameters explain the difference between trends?”. Seven parameters show considerable differences between the northern and southern provinces: Nitrogen deposition, maximum annual temperatures, mean annual precipitation, NOx emissions, Nitrogen from fertilizer, buffered soils and
ammonia emissions.
Maximum annual temperatures, mean annual precipitation, NOx emissions, nitrogen from fertilizer, and
ammonia emissions are higher in the northern provinces. And among these parameters maximum annual temperature, mean annual precipitation, and NOx emissions have a significant influence on the abundance of the
individuals living in the northern provinces. The data on the parameters maximum annual temperature and mean annual precipitation dates over ten years ago, which does not make the results reliable. Contrarily, higher NOx
emissions in the northern provinces could explain the heavier decrease of P. tithonus in the north.
However, as mentioned before, these results should be taken into careful consideration when drawing a conclusion, because other factors can influence the concentration of NOx emission in the atmosphere.
Nitrogen deposition and the percentage of buffered soils are higher in the southern provinces. And both parameters show to have a significant influence on the abundance of the individuals living in the southern provinces. Even though, the southern provinces do contain more buffered soils than the northern provinces, they still contain more unbuffered soils (63.5%). Therefore, the detrimental effects of nitrogen deposition on vegetation and butterfly species are not completely compensated by the buffering components of the buffered soils in the southern provinces. This, therefore, suggests that P. tithonus is not significantly affected by nitrogen deposition in the southern provinces, given that nitrogen deposition is higher in the southern provinces, but does not cause a heavier decline. However, to draw a real conclusion concerning the effect of nitrogen deposition on the southern individuals of P. tithonus further research is necessary.
Thus, in conclusion the following parameters negatively affect the total population of P. tithonus in the Netherlands: acoustic noise, productivity, nitrate concentration in groundwater, mean annual precipitation, number of dry days in the year, and ammonia emissions. The following parameters positively affect the total population of P. tithonus: Low values of soil nitrogen from fertilizer, buffered soils, high percentages of low vegetation, and the presence of shrubs.
When searching for an explanation for the heavier decline of P. tithonus in the northern provinces only one parameter, that affects butterflies negatively, turned out to be higher in the north and had significant influence on the abundance in that area. The parameter that could, therefore, explain the heavier decline of P. tithonus in the north is NOx emission. However, the parameter NOx emission does not have significant influence on the abundance of the total
population of P. tithonus. And this parameter is not the only parameter that differs in significance when looking at the northern, southern and total population abundance. Noticeably less parameters affect the abundance of the northern individuals (7 parameters), compared to the amount of parameters that affect the abundance of the southern individuals and the total population (13 and 10 parameters). Therefore, further research into the differences between the individuals living in the northern and southern provinces should be conducted. The contrasting results could be explained by genetic differences between the northern and the southern individuals, which would make them two different populations. However, the results of this analysis alone cannot support this conclusion.
This research, therefore, motivates the importance of further research into this specific species in the Netherlands, especially genetic research on the northern and southern individuals is recommended. Moreover, it motivates the need for further research into the changing population dynamics of widespread butterfly species, because these types of species are still declining in the Netherlands and too much knowledge on the explanation for these declines is still lacking.
Limitations of the results
My investigation has some limitations. The monitoring data of P. tithonus contained many missing values. Trim predicts the missing values by estimating the observed counts. These estimates are not objective observations, therefore some of the interpolations can differ from reality. However, trim takes into account many statistical faults that can occur to maximize the representability of the interpolated results.
All the maps that were used as covariable data represented the data of the Netherlands for one particular year. The overall population trend of P. tithonus was calculated for the period 1992 up to 2019. The trends of the
parameters over the full period of 1992 up to 2019 were not taken into account. This is because the data about the parameters over this total period could not be obtained, due to time restraints or a lack of data in public databases.
Moreover, the data about the parameters acoustic noise, productivity and weather contained information from more than five years ago (Atlas living environment, 2017; RIVM, 2013; KNMI, 2012; KNMI, 2017; KNMI, 2017). However, for the acoustic noise covariable, this was only for the noise that comes from aviation. For the data of CO2
capture which was used for the productivity parameter, more recent data should be used in further research to give a more accurate analysis. And for the weather covariables, these datasets contained information on the average value of a period of ten years, which makes this dataset more reliable, even though the data are older.
The specificity of the covariables was reduced due to a few factors. First of all, the covariables were binarily categorized. Consequently, values that were slightly above the median, were sorted in the same category as values that were significantly higher than the median. Secondly, the maps about NOx emissions, ammonia emissions, and
nitrogen from fertilizer were not observations per site, but per municipality. And lastly, the datasets of the covariables, acoustic noise, productivity, percentage of trees, percentage of shrubs and percentage of low vegetation contained NA’s. The sites that contained NA’s (i.e. not available data) could not be used in the analysis. Nevertheless, the datasets were very large, in combination with the P. tithonus dataset, because they contained information for more than one hundred sites. This compensates for the possible basality of the covariables.
Recommendations
The following recommendations were drafted for further research.
● Some of the parameter data, covered outdated information from many years ago. This research should, therefore, be repeated in a few years. Then, the monitoring data on P. tithonus will be enlarged, which would make the analysis more reliable. Moreover, new maps on the parameters may be available to make the data, used for the analysis, more recent.
● The correlations between parameters should be tested. Some of the parameters used in this analyses might be reinforcing the effect other parameters have on P. tithonus and that is information that should be taken into account.
● The trends of the individual parameters should be taken into account. In this research, data of only one year was used. However, the analysis would be more reliable if the effect of the trends of the parameters would be tested with the trend of P. tithonus.
● The effect of flower diversity and herbaceous plant diversity on this species is not analyzed in this research, due to the inability to obtain usable data. However, the effect of these parameters is of great significance to butterfly species (Kithara et al., 2008; Salgado et al., 2020). Further research should, therefore, take these parameters into account.
● Genetic research should be conducted on the individuals living in the northern and the southern provinces, to test whether the individuals are different populations and are therefore are affected differently by the parameters.
