Age structure and isolation impact the assemblage of ground beetle (Coleoptera:Carabidae) species of forests

(1)

Age structure and isolation impact the

assemblage of ground beetle

(Coleoptera:Carabidae) species assemblages of

forests

Author: K. L. de Waart

Supervisor: B. Brugge

University of Amsterdam

Faculty of Natural Sciences, Mathematics and Informatics July 22, 2017

ABSTRACT

Human activities in Northern and Western Europe have a large effect on forests. Large continuous areas become fragmentized, which could lead to the formation of biogeographic islands, and a loss of older aged stands is recorded. This research examines the effect of isolation and change in age structure of forests on the ground beetle (Coleoptera:Carabidae) species assemblage. To determine whether isolated forests do truly form biogeographic islands and to examine the effect of age structure, the species composition of two isolated and young forests was sampled and compared with data previously obtained from an older and more connected stand. There was found that isolation and age have an impact on ground beetle species composition even though other environmental conditions may also play a role. Moreover, diversity in younger stands was higher although rarefaction analysis indicated that an increase in sampling effort could affect these findings. Additionally, one young forest stand was a predictor of less species of conservational concern. The same forest predicted more winged species while the older forest provided habitat to more un-winged species. This is an indication that the isolated forests do indeed form biogeographic islands in the landscape even though no effect was found on preferred spreading habitat type or un-winged specialist species. The same younger stand contained more generalist species; this is assumed to be due to the younger age. There is concluded that fragmentation and loss of older aged stands have an impact on ground beetle species in the Netherlands which could lead to the loss of certain species when current practices continue.

(2)

1. Introduction

Biodiversity loss is a major problem worldwide (Rockström, 2009). Decrease in species richness is for the largest part caused by fragmentation of the landscape (Rainio & Niemelä, 2003). Anthropogenic activities such as farming, forestry and urbanization cause large continuous areas to fragmentize into smaller patches (Riley & Browne, 2011). Instead of a contiguous landscape with soft transition zones in between different types of habitats, a mosaic of habitats with barely any transition develops within intensively used areas.

European forests are also affected by fragmentation; this is mainly due to forest management and agriculture. Especially Northern and Western European forests suffer from high rates of fragmentation and although the total acreage of forest is growing each year, a large part does not provide habitat to many species because of the low connectivity between different areas (Estreguil et al., 2013). These forests can therefore be considered biogeographic islands surrounded by natural and artificial landscapes completely different from the forest itself (MacArthur & Wilson, 1967) and a low connectivity with the ‘mainland’ and other islands, or in this case, other forests. When these biogeographic islands form in the landscape, migration of species in between the patches will occur less frequently. Moreover, colonizing an isolated patch will not be possible for species that cannot cross the matrix.

Besides that, there is a lack of contiguous temperate broadleaf forests older than 200 years in Europe due to forestry (Estreguil et al., 2013; Koivula, Kukkonen & Niemelä, 2002). Still, it is not completely understood to what extent the age structure plays a role in habitat quality and whether these assumedly isolated forests truly form biogeographic islands (Magura, Ködöböcz & Tóthmérész, 2001).

Fragmentation, low connectivity and younger age structure plays a major role in forests found in the Netherlands as well. Around 60% of forest is fragmented and there is barely any connectivity to be found in between different patches (Estreguil et al., 2013). Because of this low connectivity these isolated forests could be considered biogeographic islands.

The species composition of ground beetles (Coleoptera:Carabidae) has been used during previous research to determine the extent in which forest age structure influences habitat quality and whether isolated forests form biogeographic islands (Magura, Ködöböcz & Tóthmérész, 2001; Koivula, Kukkonen & Niemelä, 2002). Ground beetles are a very well-known and broadly studied taxonomic group. Some species are very vulnerable to changes in the environment and are therefore widely used as bio-indicators; their presence can give an indication of ecosystem quality and biodiversity present within a patch (Rainio & Niemelä, 2003). Fragmentation and change in age structure of forests can therefore be recognized in distribution and occurrence of certain species (Riley & Browne, 2011; Raino & Niemelä, 2003). Determining their species composition can shed a light on the way in which most migration occurs (flying, walking, rolling, floating) and thus to what extent the a forest patch is truly isolated. Moreover, all species have been classified based on their habitat preference and specialism by Turin (2000) and

Lindroth (1969). This could give an indication of the groups of species that are able to thrive in the sampled location and therefore shed a light on environmental conditions such as isolation and age structure (Lövei et al., 2006).

Several researches have been examining the effect of isolation or age structure on ground beetle communities (Magura et al., 2001; Gurdabeke et al. 2003; Rile & Browne, 2011) but, the extent to which forest species thrive in forests remains unclear (Niemelä et al., 2007) and no forests in the Netherlands

(4)

have been examined yet. It is important to understand the effects of lowering of structural forest age and fragmentation in the Netherlands to be able to foresee future impact on species that will arise when current trends in landscape change continue. Such knowledge could help to conserve and protect specific threatened forest carabid species in the future as these suffer most from anthropogenic impacts in the landscape (Koivula, Kukkonen & Niemelä, 2002). In Finland for example, several specialized species of Carabids are threatened with extinction due to forestry causing the landscape to fragmentize and forests to vanish (Rainio & Niemelä, 2003). Moreover, this research could shed a light on the impact of

anthropogenic activities on forest species as ground beetles can be used as bio-indicators.

The aim of this research is therefore to assess and compare ground beetle species compositions of two young and isolated forests (the Purmerendse Bos and the Amsterdamse Bos) with that of an older and more connected forest (the Eyserbos) in the Netherlands. While the two isolated forests are completely surrounded by rural and urban area, the Eyserbos is connected to other forests by man-made structures. From the assemblage of carabid species can be determined whether the isolated forests form

biogeographic islands in the landscape and the possible impact of age structure on ground beetle communities can be assessed.

Because the Amsterdamse Bos and Purmerendse Bos correspond in environmental conditions as they are both isolated and have a similar age, there is expected that these patches are similar in (most abundant) ground beetle species composition. The Eyserbos on the other hand is much older and less isolated; for that reason, a higher variety in species composition is expected between this forest and the other two. Moreover, there is be expected that the species found here will reflect the conditions of these forests. These expectations are based on available ecological and biological knowledge about species of ground beetles, their function as bio-indicators and the possible impact of isolation and age structure on their occurrence.

Carabids can be subdivided and classified into different groups based on their habitat preferences and distribution (Turin, 2000). Thereby, each species is grouped according to its’ eurytopicity. Eurytopic species are generalists that occur in a broad range of habitats; on the other side, ground beetles that require very specific abiotic and biotic factors are considered stenotopic species. The latter can only be found in one or a few microhabitats (Rainio & Niemelä, 2003). Additionally, species can be grouped in between eurytopic and stenotopic as well (Boeken et. al, 2002). Besides that, ground beetles can be classified according to their ability to fly. Some species are decent flyers with completely developed wings (macropterous) (Turin & Nieukerken, 2000). Because the formation of wings is energetically costly (Riley & Browne, 2011), some species have no wings (brachypterous) or only a part of the population does (dimorphic) (Turin & Nieukerken, 2000). Characteristics such as eurytopicity, flight ability and classification of species thriving in forests can be used to determine the environmental conditions within a sampled area.

Young and isolated forests are expected to show a lower frequency of stenotopic species. This is thought because specialist species suffer from a decrease of forest age structure. Koivula, Kukkonen & Niemelä (2002) found that younger forests provide shelter to mostly generalist and common species while specialist (forest) species are more abundant in older forests. Because the Eyserbos is older, more

stenotopic and forest specialist species are expected to thrive here.

Even though older forests provide habitat to more specialist species (Koivula, Kukkonen & Niemelä, 2002), this does not imply that the diversity in species and the total abundance are higher in old

(5)

____________________________________________________________________________________ forests. On the contrary, the diversity and richness in younger forests are higher than that of older forests because of the instability of ecological succession (Riley & Browne, 2011). Thereby, the total abundance is lower in older forests due to the high frequency in which generalist species are found in younger forests (Koivula, Kukkonen & Niemelä, 2002). The same pattern is expected for the forests analysed in this research. The fraction of stenotopic species found in the Purmerendse Bos and Amsterdamse Bos is assumedly lower than that of the Eyserbos, even though the richness and total abundance in species will be higher. Additionally there is expected that the Eyserbos, because it will provide habitat to more specialised forest species which have a narrower distribution and lower abundance, will have a higher degree of species that are endangered or whose appearance is not very common.

As previously state could colonization of biogeographic islands be more complicated to some species due to the hostile matrix. This also accounts for Carabids; when connectivity is limited, walking species will have problems colonizing an area even though the habitat it aims to colonize is suitable (Boeken et. al, 2002). If the Purmerendse Bos and Amsterdamse Bos could indeed be considered as biogeographic islands, less colonization is expected by brachypterous species due to the hostile surroundings compared to the that of the Eyserbos. Another indication of these two forests being biogeographic islands could be detected from the habitat in which species are able to spread to other patches. It is expected that species found here have a broad range of possible spreading habitats and thus can spread through almost all matrix surroundings. In the Eyserbos, a larger proportion of species will be un-winged and have specific demands towards the habitat type in which they can spread as there is a higher degree of connectivity with other forest patches.

Especially species that are to some extend stenotopic are expected to suffer from the hostile matrix surrounding the forests. Because they have a narrow range of habitat requirements these carabids are more vulnerable to the habitat of the matrix. This accounts for many specialist species because they are often brachypterous (Raionio & Niemelä, 2003). Consequently, stenotopic species without wings are slower and less suitable dispersers than generalists (with or without wings) (Boeken et. al, 2002) and suffer more from fragmentation. Stenotopic brachypterous species are therefore expected to be absent in the two isolated forests considering the hostile matrix. Walking specialist species could be able to colonize the Eyserbos by making use of the previously mentioned man-made structures in the landscape that improve the connectivity with other forest patches.

This research will find out which species of carabids are able to colonize relatively young, isolated forests, in particular the Amsterdamse Bos and the Purmerendse Bos, and thus whether these forests contribute to maintaining and protecting populations of Coleoptera:Carabidae species in the Netherlands. The objectives are (1) to recognize the three different clusters from the different series taken in each forest, showing that there is more similarity between series of the same forest than from different forests, and analyse the correspondence between forests (2) to determine which forest has the highest species diversity and abundance of ground beetles (3) to evaluate which most abundant species are similar or different between the forests, and what this difference between the forests implies based on ecological and biological knowledge about these species (4) to find a difference in occurrence of flight capability, habitat type of spreading, eurytopicity and brachypterous stenotopic or forest specialist species in the three forest stands (5) investigate which forest provides habitat to most species with restricted abundance, narrow distribution or which are endangered, and thus whether these forests contribute to maintaining and protecting species of carabids in the Netherlands.

(6)

2. Materials and methods

2.1 Sampling locations

The carabid fauna of two different areas will be sampled during this research; that of the Amsterdamse Bos, located on the South of Amsterdam, and that of the Purmerendse Bos located on the Eastern side of Purmerend. Data from a third forest, the Eyserbos located west of the village Eys, will be used during analysis.

The Amsterdamse Bos is moderately youngas plantation started in 1934. Since the start of the

construction frequent management has been taken place (Gemeente Amsterdam, 2010). On the Northern, Eastern and Southern sides of the forest mostly urban surroundings can be found. Since plantation, this area has been transformed from polder land to mainly buildings and infrastructure. The airstrip of Schiphol Airport can be found on the Western side. The Purmerendse Bos is a young forest of which construction started 30 years ago. The forest is mostly surrounded by rural area except for the Western side, where the city of Purmerend is located. Because of the differentiated landscape surrounding these two forests, the Amsterdamse Bos and Purmerendse Bos could be considered biogeographic islands. Moreover, the nearest natural forests, which could function as a source of species that can colonize these

new forests, are located dozens of kilometres away with canals in between that could function as a barrier. The Eyserbos on the other hand is a remnant of forests older than hundreds of years and less isolated. The forest is completely surrounded by rural landscape and connectivity with other patches is enhanced by construction of landscape elements such as terraces and dead-wooden walls functioning as stepping stones and corridors.

2.2 Sampling ground beetle fauna

During this research the carabid fauna assemblage of two forests was determined. Throughout the forests series (locations within the forests) were chosen to place pitfall traps. Each series consisted of five plastic cups that were dug into the soil in such a way that they were flush with the surface. Each cup was located five meters away from the subsequent

one. The cups were filled with formaldehyde (diluted water 1: 10) as conservative and a small amount of soap in order to decrease water tension and thus let the organisms submerge. Afterwards the trap was covered with a wooden plate attached onto the soil with nails to protect it from rain and damage. The plates were covered up with plant material as camouflage. A small opening in between the soil and the plate was left so there was space for soil fauna to crawl into the cup (Picture 1). Because carabids are often

Picture 1: Set up pitfall trap. A plastic cup was dug into the soil

and covered with a wooden plate. Branches were placed in between the plate and the soil so the trap was still reachable.

(7)

____________________________________________________________________________________ dispersed throughout an area in small populations instead of being homologous spread (Raino & Niemelä, 2003) a diversity of locations was chosen. Therefore biotic and abiotic conditions were recorded (soil, light invasion, litter and dominant vegetation) to select the most diverse sites.

Nine series in the Amsterdamse Bos and eleven series in the Purmerendse Bos were placed. These forests were sampled for a time span of 63 days. Because species have a variety in activity throughout the different seasons (Rainio & Niemelä, 2003) it is important to sample during the whole year. Due to a time limit, catching was done during the broadest time scale possible; three times with time intervals of three weeks in from March until June, the period in which most species of ground beetles are active. When traps were emptied the formaldehyde was refreshed. After the third time all of the traps were removed and holes filled up with soil.

The content of emptied traps was washed with water and afterwards the ground beetles were selected and preserved in 95% ethanol. The ground beetles found were identified by making use of “De Loopkevers van Nederland & Vlaanderen” by Boeken, Desender, Drost, van Gijzen, Koese, Muilwijk, Turin & Vermeulen (2002.

For each series the quantity of caught individuals was recorded. From the Eyserbos, data collected by supervisor B. Brugge in the years from 2012 to 2014 was used for analysis; during this research the same catching methods were used but the time scale was different. For four years, five series of pitfall traps were placed for one week halfway of June thus for a total of 28 days. Data of the species composition and the ecological characteristics and classification of the three forests was collected. Following

classifications and ecological characters of the species that were used for analysis were documented: the status of the species in the Netherlands, Belgium, Denmark and Luxemburg, the status of the species in the Netherlands, the distribution in the Netherlands, how important the species’ population in the Netherlands is in its distribution in Europe (so called I-species), which type of habitats a species can migrate through to spread to other habitat patches, the classification of the species by Lindroth (1969), the degree of eurytopicity of the species and the flight capabilities of the species. The possible inputs for each characteristic can be found in Table 1.

(8)

Characteristic Possible inputs and meaning Reference Status of the species in the

Netherlands

“in” = species abundance increases

“tin” = species abundance has a tendency to increase, “nc” = no change in species abundance

“1” = a tendency to decrease in abundance “2” = clear decrease in abundance

“3” = strong decrease in main part of the area

“4” = strong decrease in abundance in the whole area “0” = status in this area is not known)

Turin (1990)

Status of the species in the Netherlands

“VC” = very common “C” = common

“AC” = averagely common “NC” = not common “AR” = averagely rare “R” = rare

“CR” = common but only in restricted areas

Turin (1990)

I-species “NI” = population in the Netherlands not of international importance “2”/”3”/”4” = population Netherlands of international importance

Siepel et al. (1993)

Habitat spreading type “1” = spreading by following in the coast “2” = spreading in low moor and sea clay areas “3” = spreading in the surroundings of rivers

“4” = spreading in the South of the Province of Limburg “5” = spreading in South- or Southeast of the Netherlands “6” = spreading in the East of the Netherlands

“7” = spreading on higher grounds “8” = spreading on sandy soils “9” = spreading in forests

“0” = spreading throughout the whole of the Netherlands “R” = rest group which spreads without a clear pattern

Boeken et al. (2002)

Classification Lindroth (1969)

“A” = arboricol (living in trees) “X” = xerophyle (drought loving) “H” = hydrophyle (humidity loving)

“N” = mesophyle (inclusing eurytopic species) “W” = forest species

“1” = strong “2” = average “3” = bound to coast

Eurytopicity Ranging from 1 (stenotopic) to 10 (eurytopic) Boeken et al. (2002) Flight capabilities “brch/BRCH” = brachypterous species being unable to fly

“macr/MACR” = macropterous species that have fully developed wings “dim/DIM” = dimorphic species of which some carry wings and others do not have wings

“poly/POLY” = all transitions in between brachypterous and macropterous can occur

Table 1: Possible inputs for each ecological characteristic recorded for the found ground beetle species. These inputs were

(9)

____________________________________________________________________________________

2.3 Data analysis

Detrended correspondence analysis was conducted to compare the similarity in species

composition between all the series sampled throughout the forests. DCA calculates the multidimensional distance between data points, which are in this case the series within the forests. This was needed to control whether series were representative for the species composition within that forest. The distances between series of the same forests were supposed to be lower than the distances between series of

different forests in order to be able to treat them as one forest. When this showed to be true, species found in series from the same forest could be merged and further analysis was conducted on this combined data. Moreover, the correspondence in species composition of forests could be compared.

To find out which forest had the highest species diversity the Shannon-diversity index (Shannon & Weaver, 1998) was used. This measurement takes the total species count and the evenness of species distribution into consideration. The forest with the highest Shannon Index H value was considered to be the forest with the highest diversity in carabid fauna. Shannon indexes cannot be compared proportionally between the forests because the Shannon index does not increase proportionally when diversity does. Thereby, Shannon diversity index calculations do not take a difference in sampling effort and sample size into account. To investigate what expected proportion of species has been found and whether Shannon index values can be considered trustable, a rarefaction analysis (Foote, 1992) has been conducted. This analysis calculates the richness of (sub)samples. It randomly selects subsamples from the data to determine whether a new undiscovered species has been detected. Samples with a higher proportion of singletons are therefore expected to have a higher found diversity when sampling effort increases. The rarefaction curve shows how much of the expected amount of species have been found considering the sample size. The asymptote of each curve displays the total expected species on the location. The more species are expected to have been found, the closer the curve that rarefaction displays approaches the asymptote.

The abundance of ground beetles was calculated from the total of individuals caught. These values were only an indication of the total abundance of ground beetles as it has to be taken into account that the catching rate is different for each species and the sampling effort was different. Therefore, only the Eyserbos, which was sampled with the least effort, could be compared to the Amsterdamse Bos and Purmerendse Bos. Additionally, the Amsterdamse Bos could be compared to the Purmerendse Bos because less series were placed on this location.

Dominant species were selected based on the relative abundance of each species in the forest. For each forest the relative abundance of each species was calculated and visualized with a dot plot. Species with a relative abundance from 5% (rounded) or higher were considered dominant species. Because of their function as bio-indicators, the dominant species found in each forest can be used to assess environmental conditions in and around the forests. This was done by making use of the book ‘De Nederlandse

Loopkevers - Verspreiding en Oecologie’ by Hans Turin (2000) and other available literature. To assess which forest provides habitat to most species that have a narrow distribution range, a low abundance or are threatened, logistic regression (McCullagh & Nelder, 1989) was conducted. There was analysed whether the location in which there was sampled had an impact on the occurrence of species belonging to one of the previously mentioned groups. To do so, the species were subdivided into groups; one

consisting of species that are of importance to conservation, and one of species that are widely distributed. From several characteristics in the dataset these species were selected. Species of which the status in the

(10)

Netherlands, Belgium, Denmark and Luxembourg is decreasing or have a tendency to increase, were all assigned to the species relevant to conservation. Additionally, species that are decreasing in abundance or that are (averagely) rare in the Netherlands were also designated to this group. At last, species of which populations in the Netherlands are of international importance were selected. The species that do not have any of these characteristics were assigned to the group of less importance for conservational purposes.

In the DCA plot, visualising the series, species were added. This showed the distribution of species over the forests and how the species contributed to the variance in series’ species composition. Species’ characteristics (flight capability, habitat type of spreading, eurytopicity, stenotopic brachypterous species and forest specialist species) were displayed in the DCA plot to be able to detect a trend in differences of the occurrence and distribution of analysed characteristics over the forests. Logistic regression

(McCullagh & Nelder, 1989) was used to assess whether the location determined the occurrence of these species’ characteristics.

To assess the impact of location on the proportion of winged and un-winged species, multiple logistic regressions were conducted. This was done because dimorphic and polymorphic cannot be classified by winged or un-winged and thus their way of migrating is not known. For the first regression model, polymorphic and dimorphic species were considered as winged species (macropterous). The second model considered these classes to be un-winged species and during the last they were completely excluded from analysis. A binary value of 1was given to species that were macropterous and a value of 0 to brachypterous species. With logistic regression (McCullagh & Nelder, 1989) there was determined whether the location had a predictive character on the flight capabilities of the species found during sampling.

The habitat type in which species were able to spread (previously discussed) was subdivided into two groups; species having a specialised spreading habitat type (numbers 1 - 9 in the dataset) or having no specific need in habitat for dispersal (number 0 in the data set). The rest group (“R” in the dataset) was excluded because no clear information is known about spreading habitat types of these species. Generalist spreading species were given a binary value of 0 and species that are specialists in spreading habitats were given a value of 1. There was examined whether the forests could predict the species’ spreading habitat types by making use of logistic regression (McCullagh & Nelder, 1989).

To analyse whether there is a difference in eurytopicity of species among the forests, species were subdivided into stenotopic species (eurytopicity of 1 - 4) and eurytopic species (eurytopicity of 7 - 10). Species with a eurytopicity of 5 or 6 were excluded because they are considered of neutral ecological value. Stenotopic species were given a binary value of 1 and eurytopic species a value of 0. Logistic regression (McCullagh & Nelder, 1989) was conducted to determine whether the location had was a predictor of eurytopicity of the found species. The same analysis was done for forest species classified by Lindroth (1969). Forest species were given a binary value of 1 and the other species a value of 0. With logistic regression (McCullagh & Nelder, 1989) there was checked whether forests had an impact on the occurrence of forest species.

There was searched for a difference in abundance of brachypterous and stenotopic species between the forests. To do so, only brachypterous species with a eurytopicity ranging from 1 to 4 were selected and a binary value of 1 was given. All species that did not belong to this group were given a value of 0. Subsequently, logistic regression (McCullagh & Nelder, 1989) was used to find out whether the location were predictors of the occurrence of this group of species.

(11)

____________________________________________________________________________________

3. Results

3.1 Similarity in species composition of series within each forest

The species composition of series within each forest showed a higher correspondence than assemblages of series between forests. This can be seen from the three clusters that are clearly detectable in Figure 1 and the fact that most of the explained variance can be found on the DCA1 axis on which forests are

separated.

Furthermore, the similarity in species composition between the Amsterdamse Bos and Purmerendse Bos was higher than these compared to the Eyserbos (Figure 1). This means that these two forests have more species in common.

Figure 1: Outcome of detrended correspondence analysis of species composition for each series. Axis DCA1 and DCA2

show the distance between data points (series) in a multidimensional cloud. The explained variance shows what fraction of the total variance each axis reflects. The DCA1 axis, on which the forests are clustered, explains most of the variance (52.9%) compared to DCA2 (30.4%). Moreover, the DCA1 axis shows most of the variance in series between forests; the three clusters are clearly detectable. The Amsterdamse Bos and Purmerendse Bos show the highest correspondence as these cluster closely together. The Eyserbos on the other hand is clustered on the other side of the DCA1 axis and is therefore more distinct in species composition.

3.2 Species diversity, richness and abundance

Altogether, 23 species were found and 746 ground beetles caught in the Purmerendse Bos. Shannon H-index was found to be 1.876. In the Amsterdamse Bos, 19 different species and a total of 324 ground beetles were found in the nine pitfall traps and the calculated H-index was 1.715. In the Eyserbos, a total of 28 species and 1,000 individuals were caught within the 28 days of sampling with five pitfall traps.

(12)

This forest had a Shannon diversity value of 1.683. Rarefaction analysis showed the interpretation of Shannon indexes have to be treated cautiously because an increase in sampling effort could have an effect on the outcome. The Eyserbos had the highest richness in ground beetle species (Figure 2). The

Amsterdamse Bos was found to be least species’ rich. The curves in Figure 2 show that from all the species occurring in the Eyserbos and Purmerendse Bos most have been caught during sampling. The Amsterdamse Bos on the other hand is expected to provide habitat to many more species that have not been discovered during sampling. This means that diversity in the Amsterdamse Bos would be higher than that of the Purmerendse Bos and Eyserbos when sampling effort increased.

Figure 2: Outcome of the rarefaction analysis of the species diversity of the three forests. The y-axis shows the mean species

richness of subsamples during rarefaction analysis, while the x-axis reflects the size in individuals of the subsamples. When comparing the species richness of the total catch sampled, the Eyserbos has the highest richness and the Amsterdamse Bos the lowest. The effect of an increase in sampling effort can also be detected in this figure. While the curves of the Eyserbos and Purmerendse Bos are reaching the asymptote, the curve of the Amsterdamse Bos is still ascending. An increase in sampling effort will thus lead to the findings of more species in this location; therefore, the found diversity in the Amsterdamse Bos could become higher than that of the Eyserbos.

3.3 Most abundant species

The dominant species found in the Purmerendse Bos were Pterostichus strenuus, Carabus nemoralis,

Loricera pilicornis, Limodromus assimilis and Pterostichus oblongopunctatus (Figure 3). Besides P. oblongopunctatus the same species were dominant in the Amsterdamse Bos. Other dominant species

(13)

____________________________________________________________________________________ species in the Eyserbos were Abax parallelelepipedus, Pterostichus oblongopunctatus, Trichotichnus

nitens, Carabus violaceus purpurascens and Pterostichus madidus (Figure 5).

Figure 3: A dot plot of the relative abundance of species found in the Purmerendse Bos. Names of the species that had an

abundance of more than ± 5% are displayed. These are considered the dominant species. In the Purmerendse Bos, these species were Loricera Pilicornis, Pterostichus strenuus, Nebria brevicollis, Carabus nemoralis, Limodromus assimilis and Pterostichus

oblongopuncttatus

Figure 4: A dotplot of the relative abundance of species found in the Amsterdamse Bos. Names of the species that had an

abundance of more than ± 5% are displayed. The dominant species in the Amsterdamse Bos were Pterostichus albipes, Carabus

nemoralis, Loricera pilicornis, Notiophilus biguttatus, Limodromus assimilis and Nebria brevicollis.

Figure 5: A dot plot of the relative abundance of species found in the Eyserbos. Names of the species that had an abundance of more than ± 5% are displayed. The dominant species in this location were Pterostichus madidus, Carabus violaceus

(14)

3.4 Species of conservational concern

Binary logistic regression showed that the Amsterdamse Bos had less species that have a low abundance in the Netherlands and/or Europe or which are endangered (Estimate = -1.3218, Std = 0.5627 and p = 0.0188). The other two locations did not have a difference in species of this classification (Estimate = 1.1787, Std = 0.6784 and p = 0.0823 for the Eyserbos; Estimate = 0.2803, Std = 0.7363 and p = 0.7034 for the Purmerendse Bos).

3.5 Difference in species characteristics

3.5.1 Flight capability

Only one brachypterous species was found in the Amsterdamse Bos (C. nemoralis) and two in the Purmerendse Bos (C. nemoralis and C. granulates). Most brachypterous species can be found in the Eyserbos (Figure 6).

Figure 6: DCA plot of distribution brachypterous, macropterous and di- or polymorphic species. The DCA plot shows how

species contribute to the variance found between the forests and series. The colours indicate the characteristic belonging to these species. A clear distribution of brachypterous species can be seen in the DCA plot; seven brachypterous species were important in the variance of species composition in the Eyserbos while there can be seen that only two un-winged species were found in the Amsterdamse Bos and Purmerendse Bos.

Analysis in which dimorphic and polymorphic species were excluded clarified that the Eyserbos and the Amsterdamse Bos were predictors of brachypterous and macropterous species respectively (Estimate =

(15)

____________________________________________________________________________________ 2.4849, p = 0.0170 and Std = 1.0408 for the Amsterdamse Bos; Estimate = -2.3896, p = 0.0343 and Std = 1.1288 for the Eyserbos). No significance was found for the Purmerendse Bos (Std = -0.6131, p = 0.6703 and Std = 1.2885). When dimorphic species were defined as flying species, no significant effect of the location was found on the occurrence of winged or unwinged species (Estimate = 0.5390, p= 0.257 and Std = 0.4756 for the Amsterdamse Bos; Estimate = -0.9137, p = 0.138 and Std = -0.9137 for the Eyserbos; Estimate = -0.2766, p = 0.663 and Std = 0.6349 for the Purmerendse Bos). When the

polymorphic and dimorphic species were defined as nonflying species the same was found (Estimate = -2.657e+01, p = 1 and Std = 8.394e+04 for the Amsterdamse Bos; Estimate = 2.076e-14, p = 1 and Std = 1.204e+05 for the Amsterdamse Bos; Estimate = -1.212e-28, p = 1 and Std = 1.144e+05 for the

Purmerendse Bos).

3.5.2 Spreading type

The characteristic habitat type in which species are able to spread was not unevenly distributed over the forests. All forests provide habitat to generalist spreaders and species that spread only through forest or another specific habitat type (Figure 7).

Figure 7: DCA plot of habitat type in which spreading occurs distributed over the forests. The DCA plot shows how species

contribute to the variance found between the forests and series. The colours indicate the characteristic belonging to these species. Species that are specialists in spreading habitat type (though forest or through other specific habitat) were quite equally

(16)

Regression analysisindeed showed that there was no significant effect of the location on the occurrence of habitat in which spreading occurred (Estimate = -0.3567, p = 0.469 and Std = 0.4928 for the

Amsterdamse Bos; Estimate = 0.2025, p = 0.748 and Std = 0,6306 for the Eyserbos; Estimate = -0.2029, p = 0.759 and Std = 0.6628 for the Purmerendse Bos).

3.5.3 Eurytopicity

Most species found in the forests are eurytopic species as these had a eurytopicity value higher than 5. Even though more species were found in the Eyserbos with a lower value for eurytopicity, the

interquartile range does not include true stenotopic species. However, a difference can be detected in the fact that only the Eyserbos provides habitat to very stenotopic species with a eurytopicity of 2 or 3 (Figure 8).

Figure 8: Boxplot of the eurytopicity of species found in the three forests. All observations can be recognized by grey dots.

The vertical black line indicates the median of eurytopicity found for each forest, the interquartile range is indicated by t he coloured boxes. Total range of eurytopicity is indicated by the vertical grey lines. Even though all three forests have the same median, the Eyserbos has an interquartile range that is extended towards a slightly lower value of eurytopicity. Moreover, a difference is visible in total range of eurytopicity values found. There were species found in the Eyserbos that were strongly stenotopic (with values of 2 or 3), while those found in the Amsterdamse Bos and Purmerendse Bos had a eurytopicity range starting from 4.

(17)

____________________________________________________________________________________ Visualization of the occurrence of stenotopic and eurytopic species pointed at a small difference between the forests (Figure 9); binary logistic regression analysis showed that only the Amsterdamse Bos was a predictor of a the abundance of eurytopic species (Estimate = -1.46634, p = 0.0221 and Std = 0.64051). The other forests did not have an impact on eurytopicity (Estimate = 0.08004, p = 0.9250 and Std = 0.64051 for the Purmerendse Bos; Estimate = 0.30319, p = 0.7116 and Std = 0.85015 for the Eyserbos).

Figure 9. The DCA plot shows how species contribute to the variance found between the forests and series. The colours indicate

the characteristic belonging to these species. Stenotopic species seem to be more abundant in the Eyserbos than in the other two forests.

(18)

3.5.4 Forest species

Nine specialized forest species classified by Lindroth (1969) were found in the Eyserbos, six in the Purmerendse Bos and five species in the Amsterdamse Bos (Figure 11). Binary logistic regression analysis pointed out that this difference was not significant (Estimate = -0.95551, p = 0.0694 and Std = 0.52623 for the Amsterdamse Bos; Estimate = -0.02532, p = 0.9716 and Std = 0.71140 for the

Purmerendse Bos; Estimate = 0.20830, p = 0.7537 and Std = 0.663 83 for the Eyserbos).

Figure 10: DCA plot of the distribution of forest specialist species over the three locations. The DCA plot shows how

species, of which the colour indicate whether they belong to this group, are distributed over the forests and how these contribute to the variance found between the forests and series. It is visible that more forest species were found in the Eyserbos than in the Amsterdamse Bos and Purmerendse Bos, as most belonging to this group are centred around series of the Eyserbos.

(19)

____________________________________________________________________________________

3.5.5 Brachypterous stenotopic

No brachypterous stenotopic species were found in the Amsterdamse Bos and only one species in the Purmerendse Bos. Three of such species were found in the Eyserbos. Because in total only three species belonging to this group have been documented (Figure 10) and this sample size is considered too small, logistic regression was not conducted as no significant effect could have been found.

Figure 11: DCA plot from the occurrence of brachypterous stenotopic species in the three forests. The DCA plot shows

how species, of which the colour indicate whether they belong to this group, are distributed over the forests and how these contribute to the variance found between the forests and series Two species belonging to this group were found in the Eyserbos, while only one was found in the Purmerendse Bos and none in the Amsterdamse Bos.

4. Discussion

In this paper I looked into dispersal related characteristics of ground beetle species, their habitat preferences and the distribution of these characteristics wit in three forests. By looking at the ground beetle species composition there was determined whether the age structure had an impact on occurrence, diversity and abundance of carabids. Moreover there was investigated whether the Purmerendse Bos and Amsterdamse Bos could be considered isolated biogeographic islands.

(20)

4.1 Species assemblages and diversity in the three forests

A higher similarity among the series of each forest was found than between series of different forests (Figure 1). The Amsterdamse Bos differed the most from the Eyserbos, while it had a higher resemblance with the Purmerendse Bos. Because of the similarity in age and assumedly isolation of these two forests this correspondence is an indication that these environmental factors have an impact on the ground beetle species composition. The geographical location of the forests has to be taken into consideration as the Purmerendse Bos and Amsterdamse Bos are more closely located and this could affect the similarity in species composition due to the geographical range limits of dispersing species. Moreover, these forests have been sampled during another period than the Eyserbos. To be certain whether the similarity in species composition was not influenced by timing of sampling and geographical location, future research should select multiple forests distributed throughout Netherlands and sample during the same period.

Due to a difference in sampling effort the abundance of ground beetles cannot be compared among all three forests. Even though sampling in the Eyserbos was conducted during the shortest time span and less pitfall traps were placed than in the other two forests, the total amount of carabids found was higher. This was unexpected because according to Koivula, Kukkonen & Niemela (2002) ground beetles are less abundant in older forests. The timing of sampling could play a role because species have a variety in activity peaks their abundance fluctuates throughout the season. For that reason the catching rate could differ throughout the whole season and year-around sampling would strengthen the confidence in these findings (Werner & Raffa, 2003). Another possibility is that environmental conditions such as isolation or management practicis within the forests had an impact on the abundance

Shannon index and rarefaction were used to analyse species diversity. Shannon indexes calculated for each forest indicated that the Purmerendse Bos had the highest diversity, followed by Amsterdamse Bos. The older forest, the Eyserbos, had the lowest H-value for diversity. This was expected because Riley and Browne (2011) earlier discovered that the younger the forest, the higher the diversity of ground beetle species due to the instability of regenerating ecosystems. This would imply that the Purmerendse Bos (and the Amsterdamse Bos to a lesser extent) has an unstable ecosystem in which colonization, extinction and migration of populations is occurring more often than in the older and more stable forests (Riley & Browne, 2011).

The effect of difference in sampling size on diversity measurements was visualized with a rarefaction curve. Rarefaction analysis indicated that the total sample of the Eyserbos had the highest richness in species while the Amsterdamse Bos had the lowest. This can be explained by the small sample size found in the Amsterdamse Bos. From rarefaction analysis can also be determined that when sampling effort would increase, the richness and thus diversity found in the Amsterdamse Bos is expected to overgrow that of the Purmerendse Bos and Eyserbos. An increase in sampling effort could support this expectation and strengthen the findings of diversity measurements (especially that of the Amsterdamse Bos). Thus, the calculated Shannon-indexes should be handled with care. Moreover, it should be taken into account that these calculations are only an estimation of the diversity during a short time span. When sampling all-year around, these findings could become more robust. In addition, catching rate for each species may differ (Koivula, Kukkonen & Niemel, 2002) and this will influence estimates of species composition, abundance and diversity.

(21)

____________________________________________________________________________________ Analysis of species composition of most abundant species found in each forest showed a higher similarity in composition of the Amsterdamse Bos and Purmerendse Bos. The Eyserbos seemed very distinct in common species composition from the two other forests; this was also expected from the age and similar environmental conditions such as isolation and size of the Amsterdamse Bos and

Purmerendse Bos and the geographical location of the forests. Thereby, results from DCA reinforced (Figure 1) this expectation.

Dominant species found in the Eyserbos have a similarity in their distribution; most of them are restricted to an area in which they have been observed. Its’ geographical range divers from distribution only in the South of Limburg to expansion towards the Veluwe. Hence, these species have not been found in the Amsterdamse Bos and Purmerendse Bos. On the contrary the most common species in these two forests have a much broader geographical distribution throughout the Netherlands (Turin, 2000). Another difference is that the species thriving in the Eyserbos have a more narrow range of habitat as most are restricted to forests. The dominant species in the Amsterdamse Bos and Purmerendse Bos are generalists and are thus capable to survive in a broader range of habitat types.

Abax parallelepipedus is the most frequently found species in the Eyserbos. The reason for

absence in the other two forests is clear; A. parallelepipedus is a species that is mainly found in the province of Limburg. Additionally, it navigate from fields to forests by following the darkest point on the horizon (Turin, 2000). In urban environments structures such as buildings could disturb its’ navigation. Even though there have been a few observations in the surroundings of Purmerend (Turin, 2000), it is not surprising that these species have not been able to colonize the Purmerendse Bos (yet). Other species that were bound to the Eyserbos were C. violaceus purpurascens, P. madidus and T. nitens. All three species are restricted to areas which do not overlap that of the Amsterdamse Bos and Purmerendse Bos, even though a population of C. violaceus purpurascens in Schoorl (Noord-Holland) has been found. This is probably caused by dislocation when plants were transported to this area. Although observations have been made of P. madidus in the Amsterdamse Bos (Turin, 2000) the presence of this species there is questionable. Pterostichus madidus is a stenotopic species and water is a large barrier for this species. Since P. madidus is brachypterous it is not considered a fast disperser. For that reason their presence in the Amsterdamse Bos and Purmerendse Bos has been considered unlikely by Turin (2000). Trichotichnus

nitens is a true stenotopic forest species that has only been found in the South of Limburg. Because the

Eyserbos provides habitat to previously discussed forest specialist species that can only be found in a restricted area, this forest can be considered important for the conservation of the carabid fauna in the Netherlands.

Only one dominant species, P. oblongopunctatus, was found in the Eyserbos that has also been found in the largest numbers in the Purmerendse Bos. Interestingly, this species is very abundant in the Purmerendse Bos (36.7% of total caught ground beetles belonged to this species) but no individual has been found in the Amsterdamse Bos at all. One cause of this difference could be that individuals of P.

oblongopunctatus were not able to colonize the Amsterdamse Bos because wings are often not developed

enough to be able to fly; thereby, it is known to be a slow disperser (only 20m during a whole season) (Turin, 2000). This would imply that the Amsterdamse Bos is more isolated for species that disperse walking than the Purmerendse Bos is. However, since there were species found in the Amsterdamse Bos that are brachypterous, such as C. nemoralis, this explanation may not be sufficient. Another explanation for their absence could be that N. brevicollis, which is very abundant in the Amsterdamse Bos, was one of

(22)

the first species to arrive and therefore the competition pressure on P. oblongopunctatus prevented its establishment . Because both species do co- occur in in the Purmerendse Bos further research is needed to determine the extent of the competition between these two species. Finally another explanation could be that P. oblongopunctatus suffers from toxic elements in the environment of the Amsterdamse Bos. This species can be used as an indication of environmental stress such as climate and toxic elements because larvae and adult survival reflect disturbances in the ecosystem (Bednarska & Laskowski, 2009). Its’ absence in the Amsterdamse Bos could in this case be caused by chemical pollution such as heavy metals or chemicals. Some (mixtures) of these elements cause a reduction in survival, a higher intolerance to pesticides (Gardiner & Harwood, 2017) or lower respiration rate (Bednarska & Kaszowska, 2014). Although highly hypothetical, a source of toxic elements could be the emissions or the release of kerosene from planes crossing over to or from the airport that is located nearby the Amsterdamse Bos. Another reason could be the contamination of heavy metals due to construction activities (Gardiner & Harwood, 2017) or use of pesticides in the surroundings (Bednarska & Kaszowska, 2014). Because there is a large knowledge gap about the effects of (mixtures of) different chemicals on populations of P.

oblongopunctatus, more research is needed to find out whether any of these explanations could be the

reason for their absence.

Another species that differs in abundance between the Amsterdamse Bos and the Purmerendse Bos is P. strenuus. In the Purmerendse Bos, 4.9% of total catches belonged to this species. In the

Amsterdamse Bos, this was only 0.6% even though it is a very common species in the Netherlands. As P.

strenuus was found to have a fast dispersal rate (den Boer, 1970) this difference could be caused by

interspecific or intraspecific interactions with other species or differences in forest structure. Prevailing species that the two young forests have in common are all eurytopic species such as C. nemoralis, L.

assimilis, L. pilicornis and N. brevicollis. The species L. assimilis and L. pilicornis were not found in the

Eyserbos. As L. assimilis has a competitive relationship with P. nigrita (Turin, 2000), its’ absence could be explained by the presence of P. nigrita in the Eyserbos (1.2 %). The findings of ground beetles that belong to the very common and eurytopic L. pilicornis in the two young forests could possibly be explained by their tolerance of anthropogenic impacts in these areas. This species is known to reflect disturbances in the environment such as eutrophication when found with other eurytopic species (Turin, 2000). Besides that, it is a very fast disperser by flight (Turin, 2000) and for that reason may have been one of the first species to colonize the Amsterdamse Bos and Purmerendse Bos.

Paranchus albipes was the only dominant species in the Amsterdamse Bos that is stenotopic.

Some were found in the Purmerendse Bos as well, but none in the Eyserbos. This species is macropterous although wings are strongly reduced in size. To be able to understand how these species colonized these areas, an investigation of the state of the wings of found ground beetles is needed. Paranchus albipes is a hydrophilic species that mostly thrives nearby water bodies (Turin, 2000; Neumann, Griffiths, Hoodless & Holloway). A possible explanation for the absence of this species is that sampling in the Eyserbos did not find place nearby water bodies. Finally, competition could again be an explanation.

The species Notiophilus bitgutattus has surprisingly enough only been found in the Amsterdamse Bos. Because of its’ eurytopic character, generalist spreading behaviour and high mobility, observations in the other two forests (and especially in the Purmerendse Bos) can really be expected. Therefore it is highly possible that other characteristics except for ability to fly, spreading habitat type and eurytopicity are playing part in the potential of species to migrate to other forest patches.

(23)

____________________________________________________________________________________

Carabus nemoralis and N. brevicollis were found all of the three forests, although much more

abundant in the Amsterdamse Bos and Purmerendse Bos. Both species are very eurytopic and have many habitat types in which they can occur; especially N. brevicollis is found in many locations as long as the environment is humid enough. Because of this highly eurytopic character, this species is assumed to be a good disperser as it is able to cope with a broad diversity of environmental conditions even though wing development is poor. The large number in which these were found is probably caused by the timing of this research; as the early peak of unhardened, young adults is finished pupating in spring and will be active (Turin, 2000).

4.2 Species of conservational concern

The dominant species composition strongly indicated that the Amsterdamse Bos and Purmerendse Bos provided habitat to less species that are restricted in occurrence, have a narrow distribution or which are endangered as these are all widely distributed throughout the Netherlands. Logistic regression indeed showed that the Amsterdamse Bos was a predictor of the occurrence of less species belonging to this group. Only one species was found in the young and isolated forests that met all of the criteria on which was selected; four of such species were found in the Eyserbos. This difference was expected because it has been found during previous research that ancient forests are of importance to several species that are bound to the conditions in these locations (Sroka & Finch, 2005). Because of a loss of older aged forests (Estreguil et al., 2013), species thriving in older forests such as the Eyserbos are expected to suffer from a loss of habitat. Even though the findings were not significant for the Eyserbos and Purmerendse Bos, a stronger effect is expected when more locations with a variety of age structure would be sampled. Thereby, future research should not only take age structure into consideration but also other environmental variables such as vegetation structure that could influence the occurrence of species of ground beetles (Sroka & Finch, 2015).

4.3 Impact isolation and forest age

4.3.1 Flight capability

It was found that the Amsterdamse Bos had a higher abundance of winged species while the Eyserbos had a higher ratio of un-winged species. This was expected when considering the hypothesis that the Amsterdamse Bos can be considered a biogeographic island. Thus, un-winged species are suffering from the hostile environment and therefore have a smaller chance of migrating to the Amsterdamse Bos. Non-flying species do have a chance to colonize the Eyserbos as this is more connected with other forest patches due to the landscape structures that were build; Fischer et. al (2013) found earlier that these hedges help brachypterous species to migrate.

During this research the possible effect of the time that it takes for un-winged species to colonize an area was not taken into account; therefore, it is possible that over the years more un-winged species will colonize the Amsterdamse Bos as migration of these species is slow. Moreover, species will lose their wings in older-aged forests because the ability to fly is less advantageous in an older forest (Riley & Browne, 2011). Therefore, sampling during a time-span of a couple of years and examining the found ground beetles on flight ability is needed as this could help to understand which species have been

(24)

colonizing and how species changed in wing state. This could make clear whether this found effect can with certainty be subscribed to the isolation of this forest.

My results indicate that the Amsterdamse Bos is more isolated than the Eyserbos, and could indeed be considered a biogeographic island in the landscape. The Purmerendse Bos however did not predict un-winged species as expected. For that reason it seems as if the Purmerendse Bos is less isolated than the Amsterdamse Bos. One reason could be a difference in the matrix surrounding these two forests. The Amsterdamse Bos is almost completely surrounded by infrastructure and buildings, while the Purmerendse Bos is still connected to rural landscape. An increasing urbanization gradient was found to have a positive effect on the abundance of winged species (Deichsel, 2006). Thus, due to a higher gradient of urbanization, colonizing the Amsterdamse Bos could be more difficult for walking species than spreading to the Purmerendse Bos. This is an indication that urbanization may have an effect on the degree of isolation in the Amsterdamse Bos and Purmerendse Bos.

The found effect seemed only true when polymorphic and dimorphic species were excluded; the other two models in which these groups were defined as flying or non-flying species did not show any significant results. I consider the result when poly- and dimorphic species were excluded reliable because excluding these species does not have an impact on the abundance of brachypterous and macropterous species. Because these species have a large variety in their flying abilities and can develop wings within a couple of generations (Riley & Browne, 2011) it is not known how these colonized the areas sampled during this research. Therefore, it is only possible to assess how the species arrived by looking at the wing state of found ground beetles. To be able to group dimorphic and polymorphic species according to their ability to fly, and thus include these species during analysis, future research should look at the wing state of poly- and dimorphic ground beetles and determine whether these were able to fly.

4.3.2 Spreading type

The expected difference between the forests surroundings in beetles’ preferred spreading habitat type has not been found. This indicates that species could colonize the forests irrespective of their specificity in spreading habitat, and thus that the Purmerendse Bos and Amsterdamse Bos are not isolated from perspective of these species. This is surprising because species that can only spread through forest habitats are not expected to be able to colonize the isolated forests. A reason for this unexpected result could be that the flight ability of the species was not taken into consideration during analysis. All of the species found in these two forests that had specific requirements for habitat type of spreading were polymorphic, dimorphic of macropterous and thus had in principle the possibility to fly. Therefore these species may not need a corridor to migrate but only small patches functioning as stepping stones. The brachypterous species found in these forests were all generalists in spreading type, and were thus more resistant to the hostile matrix. In the Eyserbos on the other hand, there were species found that were brachypterous and had specificity in habitat spreading type. However to establish if there indeed exists a causal relation between ability to fly, spreading habitat type and colonization, additional analysis is needed.

4.3.3 Eurytopicity

Slightly more specialist species were found in the Eyserbos. Moreover, species were found in the Eyserbos that are to a higher degree stenotopic than the species found in the other forests (Figure 8). The Amsterdamse Bos is a predictor of eurytopic species; thus, less stenotopic species can be found on this

(25)

____________________________________________________________________________________ location. This was expected because it was found during previous research that young forests provide habitat to more generalist species and older forest to more specialist species (Koivula, Kukkonen & Niemelä, 2002). Surprisingly this effect was only found for eurytopicity of species in the Amsterdamse Bos. No negative impact was found from the Purmerendse Bos on the occurrence of specialist species even though that would be expected because the Purmerendse Bos is younger and the effect is expected to be stronger.

A possible explanation is that during this research eurytopicity and specialization for a certain habitat were not differentiated. A species can be eurytopic within a specific habitat type, meaning that the requirements within this habitat are broad but occurrence in other habitat types does not occur (Turin, 2000). An example of a such a species is L. pilicornis, which is a very eurytopic species thriving in many different forest types, but outside of forests it has a very stenotopic character (Turin, 2000). Hence, there could still be an effect of forests on occurrence of overall stenotopic species and those that are only stenotopic with respect to the habitat type. Future research should take this difference into consideration.

Another surprising result is that the Eyserbos did not predict the occurrence of stenotopic species. Besides the previously discussed explanation, this outcome could also be explained by the fact that the Eyserbos may not have a sufficient size. Large areas provide a higher diversity of microclimatic conditions and will therefore meet the specific habitat requirements for stenotopic species (Halme & Niemelä, 1993; Koivula & Vermeulen, 2005). Although it is an older forest and therefore expected to provide habitat to a significant higher number of stenotopic species (Koivula, Kukkonen & Niemelä, 2003), the area may not be large enough to show a difference in abundance. Future research should take the effect of size into consideration when examining the eurytopicity of species within different forests.

4.3.4 Forest species

Logistic regression did not show a significant difference in forest species among the forests, even though the Amsterdamse Bos did seem to have a negative impact on their abundance as the p-value was

approaching significance (p = 0.0694). A stronger effect is expected especially considering the previously mentioned infrastructure around the Purmerendse Bos and the Amsterdamse Bos. It was found by

Koivula and Vermeulen (2005) that roads are a barrier to especially forest species; this should reinforce the effect on their occurrence in these forests. Therefore it is expected that when sampling effort increased and more different locations with a variety in age would be examined, this effect would be strengthened. Moreover, an as no effect was found in the Eyserbos, small area size may again have a negative effect on the occurrence of specialist forest species (Halme & Niemelä,1993; Koivula & Vermeulen, 2005).

4.3.5 Brachypterous stenotopic species

It was hypothesized that especially brachypterous stenotopic species would suffer from the isolation of the two forests that assumedly form biogeographic islands. This was expected because when the Amsterdamse Bos and Purmerendse Bos truly form biogeographic islands, these groups of species would suffer the most from the hostile surrounding.

From the DCA plot shown in Figure 10 can be seen that no brachypterous stenotopic species thrived in the Amsterdamse Bos and only one species was found in the Purmerendse Bos. Three species thrived in the Eyserbos and thus a trend is recognizable. Because of the small sample size regression was not conducted but an effect is still expected to be found when time-span of sampling is prolonged and examined locations increased. More samples for a longer time span and on more locations could lead to

(26)

the findings of more species, and could strengthen the hypothesis that forests such as the Purmerendse Bos and Amsterdamse Bos do indeed form biogeographic islands in the landscape. Moreover, the degree of specialization was not taken into consideration. As previously mentioned, species can bound to habitat types but with a diverging eurytopicity. Therefore, further analysis should take this into consideration when evaluating abundance of this group of species.

5. Conclusion

This research examined the effects of forest fragmentation and loss of older-aged forest stands on the ground beetle species assemblage. Therefore it has to be taken into consideration that anthropogenic activities in the landscape in the Netherlands have an impact on species of ground beetles. When current practices continue there could be a loss of ground beetle species with certain characteristics. Moreover, because of their function as bio-indicators, forest species belonging to other groups could be influenced as well. There was found that forest with similar conditions in age and assumedly isolation influence the species composition even though other environmental factors such as the degree of urbanization and geographical location could play a role as well. Younger forests showed to have a higher diversity in ground beetle species but an expected effect on higher eurytopicity and less forest specialist species has not been found. Additionally, one of the young forests showed to provide habitat to less species that are of conservational concern. Because there is a loss of older-aged stands there could be a decrease of more and more forest species that suffer from this change. There is an indication that fragmented forests such as the Amsterdamse Bos form biogeographic island in the landscape as less un-winged ground beetle species have been found on this location. The formation of biogeographic islands will have an impact on the migration of (ground beetle) species and thus protection or management is needed to improve connectivity between forests. More sampling locations throughout the Netherlands, an increase in sampling effort and a prolonged time-span are expected to reveal more of the expected results.

Acknowledgements

I am grateful to supervisor Ben Brugge for teaching me field work practices and sharing his knowledge on ground beetle ecology and taxonomy. I would also wish to thank Peter Roessingh for support in statistical analysis and for giving feedback during the project. Moreover, I would like to thank the

management of the Amsterdamse Bos that gave me the opportunity to use this location as a research area. I thank Emiel van Loon and Peter Assink for making time free to give me advice about statistical

analysis. And last but not least I would like to thank Jan Wormmeester to provide materials and a room on Science Park for the examination of ground beetles.

(27)

____________________________________________________________________________________

Literature

Barbosa, A. M., Fontaneto, D., Marini, L., & Pautasso, M. (2010). Is the human population a large‐scale indicator of the species richness of ground beetles?. Animal Conservation, 13(5), 432-441.

Bednarska, A. J., & Laskowski, R. (2009). Environmental conditions enhance toxicant effects in larvae of the ground beetle Pterostichus oblongopunctatus (Coleoptera: Carabidae). Environmental Pollution,

157(5), 1597-1602.

Boeken, M., Desender, K., Drost, B., Van Gijzen, T., Koese, B., Muilwijk, J., Turin, H. and Vermeulen, R., 2002. De loopkevers van Nederland en Vlaanderen (Coleoptera: Carabidae). Jeugdbondsuitgeverij,

Nederland, p.212.

Chapman, J. W., Reynolds, D. R., Smith, A. D., Riley, J. R., Telfer, M. G., & Woiwod, I. P. (2005). Mass aerial migration in the carabid beetle Notiophilus biguttatus. Ecological Entomology, 30(3), 264-272. Deichsel, R. (2006). Species change in an urban setting—ground and rove beetles (Coleoptera: Carabidae and Staphylinidae) in Berlin. Urban Ecosystems, 9(3), 161-178.

Den Boer, P. J. (1970). On the significance of dispersal power for populations of carabid-beetles (Coleoptera, Carabidae). Oecologia, 4(1), 1-28.

Estreguil, C., Caudullo, G., de Rigo, D., & San-Miguel-Ayanz, J. (2013). Forest landscape in Europe: pattern, fragmentation and connectivity. EUR Scientific and Technical Research, 25717.

Fischer, C., Schlinkert, H., Ludwig, M., Holzschuh, A., Gallé, R., Tscharntke, T., & Batáry, P. (2013). The impact of hedge-forest connectivity and microhabitat conditions on spider and carabid beetle assemblages in agricultural landscapes. Journal of insect conservation, 17(5), 1027-1038. Gemeente Amsterdam (2010). Werkverschaffing. Accessed on 16-3-2017 from

http://www.amsterdamsebos.nl/kunst-erfgoed/geschiedenis/werkverschaffing/).

Halme, E. and Niemelä, J., 1993, January. Carabid beetles in fragments of coniferous forest. Annales

Zoologici Fennici (pp. 17-30).

Jopp, F. and Reuter, H., 2005. Dispersal of carabid beetles—emergence of distribution patterns.

Ecological Modelling, 186(4), pp.389-405.

Koivula, M., Kukkonen, J. and Niemelä, J., 2002. Boreal carabid-beetle (Coleoptera, Carabidae) assemblages along the clear-cut originated succession gradient. Biodiversity and Conservation, 11(7), pp.1269-1288.

Age structure and isolation impact the assemblage of ground beetle (Coleoptera:Carabidae) species of forests