IMPACT OF DIFFERENT HABITAT PARAMETERS ON CARABID BEETLE ASSEMBLAGES IN SELECTED AREAS OF A FOREST-FIELD LANDSCAPE IN POLAND – 10 YEARS OF DATA

Long-term data on carabid beetles assemblages on differently managed study sites in forests and open areas were analysed to study the impact of selected environmental factors on the carabid assemblages of both the individual study sites over the years and the set of all study sites in selected years. Ordination separated forest stands from open areas along the first, and samples of 2011 from those of 2015 along the second axis. For study sites in forest stands in most cases, precipitation was a significant factor, especially precipitation in the year before the inventory. However, for the youngest forest site, the year of the study was most important, indicating a succession process. For study sites in open areas, both precipitation and temperature showed most often significant results. Analysing the impact of environmental factors on carabid assemblages in the full set of study sites in 2011 and 2015 revealed carbon content in the organic layer and distance from the nearest forest as significant factors. The results of the study extend our knowledge on the impact of environmental factors on the formation of carabid beetle assemblages in rural landscapes, which is essential in the framework of developing biodiversity conservation strategies.


INTRODUCTION
An essential factor for the species diversity in a given landscape, i.e. the number of species over a large area or region (gamma-diversity) (Whittaker 1972, Anderson et al. 2011, is its habitat diversity (e.g. Ryszkowski et al. 2002, Weibull et al. 2003, Purtauf et al. 2004, Hendrickx et al. 2007). However, this species diversity is influenced by environmental factors, both on the spatial level of the individual habitats as well as the overall landscape.
On the habitat level, one may ask about the meaning of environmental characteristics of the individual habitats for species diversity and if individual environmental factors have the same significance in different habitat types. Other environmental factors may influence on a larger spatial scale to a different extent the overall species diversity.
Some large-scale environmental factors (as precipitation or temperature) may affect the different habitats in a very similar way but differ from year to year. Even if many species show rather reproduced distribution patterns between years in a given landscape (Judas et al. 2002, Schwerk 2014, such factors might influence the overall species diversity. Other environmental factors, like for example soil parameters, maybe differently pronounced in the individual habitats and their impact on the species may differ from habitat to habitat. Moreover, individual habitats change due to the process of natural succession, during which also such factors change. Shape and speed of this process may strongly differ (e.g. Prach et al. 1993, Yang et al. 2005, Schwerk & Szyszko 2011. As a result of ecological succession, the species composition in a given habitat may change over time in a directed manner. The species composition in the habitat can also be influenced by landscape-related factors, like the type of neighbouring habitats or distance to the next forest or lake (e.g. Burel 1989, Fournier & Loreau 1999, Holland & Fahrig 2000, Millán de la Peña et al. 2003, Tropek et al. 2013. To understand the impact of different predictable and stochastic environmental factors on the formation of carabid beetle assemblages in rural landscapes is an essential issue because currently, we are observing a drastic decline in invertebrate fauna on a large scale, particularly demonstrated on the example of insects (e.g. Kotze & O'Hara 2003, Brooks et al. 2012, Hallmann et al. 2017, Homburg et al. 2019. Carabid beetles efficiently reflect the environmental variation and bear indicator potential at various spatial scales (Koivula 2011). Hence, they can be useful indicators in this regard.
However, to obtain robust data in order to deal with the aspects mentioned above, long-term studies are needed. One of such studies was initiated in a forest-field landscape in the west of Poland, during which various taxonomic groups were studied in open areas and forest sites of different management regime and age. In the present paper data of this long-term study on carabid beetles (Carabidae) on eight study sites carried out for ten years were analysed to study the impact of selected environmental factors on the carabid assemblages of individual study sites over the years and the set of all study sites in selected years. We wished to find out if the impact of the respective environmental parameters on carabid assemblages differs between forest stands and open areas and which of the parameters are of utmost importance for the formation of the carabid assemblages.

Study sites and field methods
The study was carried out on the research object "Krzywda" at Tuczno (western Poland, the Wałecki district) and surrounding forests (Fig. 1). This area is a heterogeneous landscape composed of different forests, agricultural and post-agricultural areas of different stages of succession, and watercourses and mires (Dymitryszyn et al. 2013). Eight study sites (five forest sites of different age and three open areas) were selected for carabid sampling (Table 1).
Carabid beetles were collected using pitfalls (Barber 1931) during the period 2009-2018. At each study site, three pitfall traps in a distance of about 50 m or more from each other were installed. Traps located in different study sites were also separated by a distance of at about 50 m or more in order to represent independent samples (e.g. Digweed et al. 1995, Hoekman et al. 2017. The traps were jars with a funnel (upper diameter 10 cm) installed above flush with the soil surface. A roof was installed a few cm above the funnel. The trapping fluid was 50 ml ethylene glycol. The trapping period covered continuously mid-May to mid-September in each year. The traps were checked for proper functioning in about two-week intervals.
Determination and nomenclature of the individuals collected were carried out according to Freude et al. (2004).

Environmental parameters
For each year of the study the following large-scale parameters were registered homogenously for all study sites: temperature (°C) and precipitation in the year of inventory (mm/year), temperature (°C) and precipitation (mm/year) in the preceding year, and the respective year itself (year of the study).
As parameters varied among the study sites, we registered the distance from the nearest forest (m) and the age of the study site (age in the respective year) as well as the carbon content (C%) in the organic layer and the carbon content (C%) in the mineral soil, which were elaborated for the years 2011 and 2015.
For the forest stands the stand age in the respective year was taken as age, whereas the open study sites were characterised with an age of 0 years (regularly-mown site with biomass removal) and 1 year respectively (irregularly-mown sites without biomass removal), based on cutting intensity and biomass treatment. Data for temperature (mean temperature for the respective year) and precipitation (total precipitation in the respective year) were taken for the closest location (Piła) from the Institute of Meteorology and Water Management -National Research Institute resources.
Values of carbon content were elaborated as follows: Soil samples were collected in the close neighbourhood of each pitfall trap in August 2011 and between October and November in 2015. On both forest sites and open areas, soil samples from mineral layers were taken from the depth of 0-5 cm. In the case of forest sites, additional material from organic layers was taken as samples. The collected soil samples were prepared for laboratory anal-

Statistical methods
For each study site and year, the carabid beetle catches of the three traps were pooled. For each of these samples, the numbers of species and individuals were calculated, resulting in one value for these two parameters for each study site per year. Accordingly, for the years 2011 and 2015 carbon values for carbon contents in the organic layer and mineral soil were In order to analyse the significance of the impact of the environmental factors on the carabid assemblages, constraint gradient analyses were conducted using Canoco for Windows 4.56 and CanoDraw for Windows 4.14 (ter Braak 1987, ter Braak & Šmilauer 2002. DCCA was first used to select the appropriate statistical model based on the longest gradient (Lepš & Šmilauer 2003) and then Redundancy Analyses (RDA) and Canonical Correspondence Analysis (CCA) were carried out.
In gradient analysis with the individual study sites over the years (2009-2018) as samples, year (year of the study), temperature (°C) and precipitation (mm/year) in the year of inventory, and temperature (°C) and precipitation (mm/year) in the preceding year were included as environmental parameters.
For the years 2011 and 2015, all study sites were included as samples. In these analyses age (age in the respective year), carbon content (C%) in the organic layer, carbon content (C%) in the mineral soil and distance from the nearest forest (m) were included as environmental parameters.
The samples of all study sites for 2011 and 2015 were together analysed with year (year of the study), temperature (°C) and precipitation (mm/year) in the year of inventory, temperature (°C) and precipitation (mm/year) in the preceding year, age (age in the respective year), carbon content (C%) in the organic layer, carbon content (C%) in the mineral soil and distance from the nearest forest (m) included as environmental parameters.
In all gradient analyses, dominance values (percentage share of the respective species in a sample) for the carabid species at the different sites were used. Redundancy Analyses (RDA) were conducted with scaling focused on inter-sample distances and no post-transformation of species scores. In the case of Canonical Correspondence Analyses (CCA) we used scaling focused on inter-sample distances and Hill's scaling. As dominance values were used, the data were not transformed in any gradient analysis. The significance of the individual environmental variables included in the respective RDA or CCA was tested applying Monte Carlo permutation tests (unrestricted, 499 permutations) with the automatic forward selection of variables (reduced model) as well as manual selection for each variable separately.
Some differences concerning distance from the nearest forest and carbon contents were revealed depending on the habitat type. The open areas were between 50 and 150 m away from forests. Differences in carbon contents were particularly due to the lack of an organic layer on the open areas. Regarding the data on temperature and precipitation, an increase in mean yearly temperature and a decrease in total precipitation over the years of the study could be observed (Appendices 2 & 3).

Individual study sites over the period 2009-2018
Concerning the forest sites (Table 2) precipitation in the preceding year (study sites 14 and 20), precipitation (study site 21) and the year of the study (study site 35) were detected as significant environmental factors when using forward selection. For the study site 16, no factor showed a significant result, but precipitation in the preceding years was closest to being significant. When using manual selection temperature was also a significant factor for study site 35.
Regarding the open areas (Table 3) both temperature (study site 17) and precipitation (study site 18) showed significant results when using forward selection. Concerning the study site 19, no factor was significant, but the temperature was closest. Using manual selection, as additional significant factors, precipitation and the year of study were detected for the study site 17 and temperature in the preceding year for the study site 18.

Set of all study sites in the years 2011, 2015 and 2011+2015
Analysing all study sites together for the year 2011 (Table 4) carbon content in the organic layer was significant when using forward selection. Using manual selection, the distance from the nearest forest was detected as an additional significant factor. The same results were obtained when analysing all study sites together for the year 2015 (Table 4).  The whole set of study sites for 2011 and 2015 together (Table 4) revealed carbon content in the organic layer and precipitation as significant factors when using forward selection. Additional significant factors were the distance from the nearest forest and the age of the study site when using manual selection. The first canonical axis of the CCA explained 38.6% of the variation in species data and 62.0% of that in the species-environment relationship. The second canonical axis explained 13.1% and 21.2%, respectively. As shown in the ordination diagram (Fig. 2), the forest sites were separated from open sites along the first ordination axis, with carbon content in the organic layer and age of the study sites positively correlated with the forest sites and distance from the nearest forest positively correlated with the open sites. Characteristic forest species as Carabus hortensis and Pterostichus niger were located close to

DISCUSSION
The results indicated that on individual forest study sites precipitation (particularly precipitation in the year before the inventory) was an important factor, whereas in open areas both precipitation and temperature showed some significant results. The more substantial influence of precipitation in forest habitats may be explained by shading due to the tree crowns. In unshaded, open habitats, however, the temperature may become a dominant factor because of the stronger impact of direct sunlight. Except for desert steppe, Tsafack et al. (2019) detected climatic factors, in particular temperature, as the most critical factors for variability in carabid species composition in grassland ecosystems in China. Tuf et al. (2012) demonstrated temperature-dependent differences in activity patterns of individual carabid beetle species between a forest and a clear cut area. In accordance with our results described above, regarding the youngest forest site (study site 35) temperature was a significant factor. However, the most important factor for this study site was the year of the study, indicating a succession process. This result corroborates earlier observations made for this study site (Schwerk & Szyszko 2012). Samples collected till 2013 were clearly separated from those elaborated after 2013, what may be connected to canopy closure. In a study on boreal carabid beetle assemblages, Koivula et al. (2002) revealed the canopy closure distinguished the open phases from the closed ones, the catches showing a drastic decrease and assemblage level change.
When comparing all study sites together, forest habitats were clearly separated from open habitats, indicated by carbon content in the organic layer and distance from the nearest forest as significant factors. This result supports the importance of habitat diversity for the formation of diversified carabid beetle assemblages and species richness in the landscape. The relevance of precipitation was supported by being a significant factor in separating the study sites in 2011 from 2015. Regarding the forest habitats, it turned out the organic layer was of very high importance since the factor carbon contents in the organic layer overruled the factor age of the study sites. For many carabid species, this layer has essential functions, for example as a refugee after habitat disturbances, as a place for foraging or for laying the eggs (e.g. van Heerdt et al. 1976, Bauer 1981, Skłodowski 2017. Koivula et al. (1999) showed a significant influence of leaf litter on carabid abundance in forests in central Finland. Carbon contents in the mineral soil did not have a significant impact. However, this result may be influenced by the history of the study sites, because all of them except study site 21 (naturally regenerated pine forest with a share of oak, beech and birch) were used as agricultural fields in the past (Dymitryszyn et al. 2013). Soil preparation practices as ploughing may have influenced the formation of the carabid assemblages (e.g. Skłodowski IMPACT OF HABITAT PARAMETERS ON CARABID BEETLE ASSEMBLAGES 2005, Kosewska et al. 2018). It should be noted that compared to other brown soils of the Palearctic region, the soils in our study area are sandy soils with a limited amount of soil organic carbon. Thus, the importance of soil organic content in our study is evident.
Although habitat diversity can be assessed as a key factor for species diversity in landscape, different predictable and stochastic environmental factors (climatic factors) may affect the carabid beetle composition on individual sites. For example, Judas et al. (2002) demonstrated that mesoscale (pitfall density of 10-50 km -2 ) variation in soil and microclimate govern spatial population structure of several carabid beetle species in topographically varied beechwood landscape. Besides, ecological succession on individual study sites is a process, which has some influence on both the environmental factors in the respective habitat and degree of habitat diversity in the landscape (Sferra et al. 2017, Hilmers et al. 2018. However, another important aspect is climate change. Climatic changes may have a direct impact on insect species regarding the range of expansion, growth, phenology or population genetics, and an indirect one due to effect on plants species (Jaworski & Hilszczański 2013, Pureswaran et al. 2018. The results of the study extend our knowledge on the impact of different predictable and stochastic environmental factors on the formation of carabid beetle assemblages in rural landscapes. Nowadays, to counteract the decline in insect diversity is a serious task. Concerning this matter, the results of our study can play an essential role in the framework of developing biodiversity conservation strategies. We assume that active management of habitats (ecological or environmental engineering) can be a helpful tool when it comes to the implementation of such strategies.