Aquatic insect community structure in urban ponds: effects of environmental variables
Johan Andersson
Degree project inbiology, Master ofscience (2years), 2014 Examensarbete ibiologi 45 hp tillmasterexamen, 2014
Biology Education Centre and Department ofEcology and Genetics, Uppsala University
Table of Contents
1. ABSTRACT ... 2
2. INTRODUCTION ... 2
3. METHODS ... 4
3.1 STUDY SITES ... 4
3.2 INSECT SAMPLING ... 5
3.3 ENVIRONMENTAL VARIABLES ... 6
3.4 VEGETATION COVER AND SHORELINE ... 6
3.5 CHEMICAL ANALYSIS OF WATER SAMPLES ... 7
3.6 LAND USE AND GIS ANALYSIS ... 7
3.7 STATISTICAL ANALYSIS ... 8
4. RESULTS ... 8
4.1 PRINCIPAL COMPONENTS ON ENVIRONMENTAL VARIABLES ... 8
4.2 INSECT RECORDS, ABUNDANCE AND DIVERSITY ... 10
4.3 INSECT ASSEMBLAGES AND POND CHARACTERISTICS ... 10
4.4 RICHNESS, ABUNDANCE AND DIVERSITY ... 12
5. DISCUSSION ... 13
5.1 THE EFFECT OF URBANIZATION ON AQUATIC INSECT COMMUNITY STRUCTURE ... 13
5.2 ENVIRONMENTAL EFFECTS ON SPECIES RICHNESS, ABUNDANCE AND DIVERSITY ... 14
5.3 SPECIES DETERMINATION – HIGHER TAXON APPROACH ... 15
5.4 NEWT PRESENCE ... 16
5.5. SPECIES OF CONSERVATION CONCERN ... 16
5.6 FINAL CONCLUSION ... 16
6. ACKNOWLEDGEMENTS ... 17
7. REFERENCES ... 17
8. APPENDICES ... 21
8.1 APPENDIX 1 – SPECIES LIST, OCCURRENCE AND FREQUENCY ... 21
8.2 APPENDIX 2 – INSECT ORDER SPECIFIC RDA BIPLOTS ... 24
8.3 APPENDIX 3 – LAND USE DETAIL SPECIFICATIONS ... 26
8.4 APPENDIX 4 – POND DESCRIPTIONS ... 28
1. Abstract
I sampled aquatic insects in 26 ponds of varying types in the urban landscape of the city of Stockholm and related insect community structure to environmental variables. I also related environmental factors to species richness, diversity and abundance of the sampled aquatic insects. A Redundancy Analysis (RDA) showed that the most important variables in explaining insect community structure was the remoteness to developed area and the amount of emergent vegetation in the ponds. Species richness increased with distance from developed area, diversity was related to floating vegetation and abundance of insects increased with distance from developed area and with higher amount of forestation and vegetation. The results of my study shows that urbanization effects divide the insect community into clusters of species that are tolerant or intolerant to effects of urbanisation. One internationally red-‐listed species, the dragonfly
Leucorrhinia pectoralis was found in five (19,2%) of the ponds. My result
suggested two important factors that should be considered when planning urban ponds. First, it is important to re-‐create varying types of ponds and include green buffer areas and second, plant colonisation should be facilitated to better mimic the natural states of ponds.
2. Introduction
The urban landscape is continually expanding along with rising population levels. The proportion of people living in urban areas is forecasted to grow from 50% in 2008, to 69% globally in 2050 (United Nations Population Division 2011). In recent years the field of conservation biology has widened its
trajectory from a view of preserving pristine ecosystems to also include areas highly influenced by human activities as important areas for nature and wildlife conservation (London Biodiversity Partnership 2001, Harrison & Davies 2002, Alvey 2006). Such areas could for example be green spaces within cities
(Goddard et al. 2010), which is defined as an undeveloped open space at least partly covered with vegetation including community gardens, cemeteries and parks that often contain water (EPA 2013).
Studies indicate that urbanization degrades biodiversity through various
processes including e.g. habitat fragmentation (Dickman 1987), land conversion (Moore 1990) and introduction of alien species (Kowarik 2008). The future projections on extinction rates are depressive and the risk of more species becoming red-‐listed is increasing with urban development (McDonald et al.
2008). However it has also been shown that suburban areas contain a large biodiversity of organisms, often higher than the rural outskirts and the central urban areas. This is due to a broad variety of different habitats (McKinney 2008).
For example, the common frog (Rana temporaria) in Great Britain has declined in rural areas, but has increased its abundance in urban areas (Carrier & Beebee 2003).
Factors important for a stable population density of species in the urban landscape differ. For some species landscape barriers hinder migration and
dispersal (Blakely et al. 2006), whereas the major concern among other species is the habitat patch quality (Angold et al. 2006). These factors might be especially important in urban areas such as large cities because of the high fragmentation and the low habitat quality. Unfortunately they are not well studied in large cities and warrant more research focusing on the relationship between biodiversity and habitat quality in these areas.
Even though ponds are not green in colour they are an important part of green spaces (Fontanarosa et al. 2013). The proportion of green-‐space associated with water can be quite large in cities (see refs in Gledhill et al. 2008). Traditionally ponds and smaller waters were filled in cities, but nowadays many ponds are often restored and new ones created (Gledhill & James 2012). These newly created ponds are used for urban drainage, nature conservation, ornament features and more (Sutherland & Hill 1995), and studies have suggested that ponds are important for human quality of life (Lees & Evans 2003).
The biodiversity in ponds is affected by many abiotic and biotic factors. For example, it has been shown that size and connectedness of the ponds affect the species composition, with different species occurring in larger ponds than in a set of smaller ponds. Interestingly, a few smaller ponds can even harbour larger microfaunal diversity than one large pond (Oertli et al. 2002). Many life history traits of aquatic insects rely on interactions with plants. The relationships between species richness of insects and plants are quite well studied and some of the more important aspects of insect-‐plant interactions are: herbivory, oviposition, predator evasion and foraging (McGaha 1952). Numerous aquatic insects are susceptible to fish predation and presence of fish has been shown to play a key role in structuring aquatic insect communities (Bendell & McNicol 1987). Additionally the local water chemistry is affecting aquatic insect composition, where e.g. pH is a critical factor during the development of the larval stages with many species having problems coping with a too acidic environment (Bell 1971).
Biodiversity in city ponds is also affected by land use/land-‐cover and perhaps even more so than in rural areas because land-‐cover changes are extreme in cities. While we have some knowledge on how land-‐cover affects biodiversity for terrestrial systems in cities, we know very little on how city ponds are affected.
For terrestrial systems Loss et al. (2009) showed that bird species richness was higher in urban cities with undeveloped patches and heterogeneous land-‐cover types. Interestingly, Tratalos et al. (2007) found that species richness of birds in British cities showed a humped shaped relationship with housing density, suggesting that land-‐cover variables do not necessarily have to show linear relationships with biodiversity variables. With regards to ponds in cities, less information is available and I am only aware of one study that focused on pond biodiversity and environmental land-‐cover variables. In that study, Goertzen &
Suhling (2013) found that sealed area (buildings and roads) was negatively associated with pond biodiversity.
Since many of the urban ponds are relatively new the ecological value of these ponds are still unknown and research identifying important factors affecting the
biodiversity of these ponds are still lacking (Williams et al. 2008). There is certainly a need to pinpoint the definitive associations between biodiversity and environmental variables as well as structural landscape features. Such
knowledge is important for city planners and the purpose of my study is to provide this knowledge.
I conducted my study in the conurbation of the Stockholm capital province. This is Sweden’s most heavily urbanized area consisting of an archipelago structure with a mosaic of islands and suburbs spreading in north-‐south direction from the city centre. The Swedish landscape has a long history of ditching and draining and the Stockholm area is no exception (Jakobsson 2013). Thus the landscape has been transformed and reduced of important wetland habitats. The present-‐
day restorations of wetlands in Sweden are mainly conducted for financial benefits and reduction of agricultural nitrogen pollution (Byström 1998). It has also been recognized as a cost effective way to reduce nitrogen emissions in urban settings, including Stockholm, where wetlands function as a sink of
reactive nitrogen reducing the loading and eutrophication of surrounding waters (Gren 1994). Both constructed and natural ponds and wetlands often contain a high biodiversity, today considered to provide important ecosystem services and amenity values in urban green spaces (Bolund & Hunhammar 1999). The
importance of ponds for urban invertebrates has been emphasized along with the importance of the naturalness of green spaces (Moore 1990).
In this study I ask which environmental factors affect the biodiversity of aquatic insect species in urban city ponds. I will also give suggestions on what can be done in urban conservation planning to create a high biodiversity in these ponds.
3. Methods
3.1 Study sites
The investigated ponds were located in the north central parts of Stockholm (59°19'N, 18°4'E). This is one of Sweden’s most highly urbanized areas and the ponds included in my studied were situated in the municipalities Järfälla, Sollentuna, Solna, Stockholm and Sundbyberg (SCB 2010).
Ponds that were included were in the size range of 2 m2 to 2 ha, using the definition of pond by Ponds Conservation (2002) and earlier used by Gledhill et al. (2008). To locate ponds, I contacted ecologist and official water
administrators employed at the municipalities. In addition, I searched for ponds on digital overview maps. Twenty-‐six ponds were included which resulted in both natural and constructed, permanent and temporary, both old (>100 years) and new (one year) ponds (Fig. 1). Some of them were constructed as city park ponds for recreational values whereas others had been built as stormwater ponds to collect surface runoff and prolong turnover from surrounding water bodies and decrease high nutrient and waste/pollutant loading. No golf course
ponds were included among the selected ponds. The invertebrate fauna of golf course ponds in the area has previously been studied by Colding et al. (2009).
Fig. 1. Pond locations in the Stockholm area. Grey areas represent in falling intensity; developed land, forest, other open area. Water area is represented by white. Black striped lines depict municipality borders and ponds locations are symbolised by white circles with black outline. See Appendix 4 for numbering and pond characteristics. Terrängkartan™ © Lantmäteriet 2010:
Permission I 2010/0058.
3.2 Insect sampling
Aquatic insects were sampled between May 15th and June 7th 2013. I restricted my sampling to four insect orders that are easily distinguishable from each other on site: Coleoptera, Hemiptera, Odonata and Trichoptera. Sampling of these orders covered species with different roles in the food web, both herbivores and predators from various functional groups. This approach has the advantage of simplifying species determination and is conventionally used in comparative studies (see Simberloff & Dayan 1991). I sampled the aquatic life stages of the selected insects orders which are inhabiting the pond and thus are exposed to the local environmental factors for an extended period of time in contrast to visiting insects i.e. larvae in Odonata and Trichoptera and larvae and adults in Coleoptera and Hemiptera. All ponds were only sampled on the day of the visits.
For collecting insects I used a bottom scoop net with a diameter of 20 cm and a
mesh size of 1,5 mm. Six samples were taken in each pond at a depth of 2-‐3 dm.
The net was swept along the bottom in opposite directions (left to right) eight times on a 1 m stretch, which constituted one sample. By using six samples I managed to cover all types of representative microhabitats along the shoreline:
e.g. soft bottom, hard bottom with and without vegetation. The sampling strategy was derived from the guidelines by the SEPA (2006). All insects were
determined to order at the pond site and insects from the four orders included in the study were preserved in 70% ethanol, stored in labelled plastic tubes and brought back to the laboratory for species determination. All other species were released back to their respective ponds.
Species determination of Coleoptera was carried out by Johannes Bergsten (Swedish Museum of Natural History). Trichoptera was species determined by Ulf Bjelke (Swedish Species Information Centre) and I determined Hemiptera and Odonata. Specimens that could not be determined to species level were still included in the final analysis and set to family or genus-‐level and hence regarded as separate taxa. In most cases these specimens were early instar larvae. Larvae of Coenagrion pulchella and C. pulchellum are not distinguishable and were therefore regarded as same species in my analysis. The same applies to three cases among the Trichoptera were larvae could not be distinguished between two species. These were i) Limnephilus affinis and L. incisus, ii) Limnephilus luridus and L. ignavus and iii) Oligotricha stricta and O. lapponica. These three species were recorded in only one pond each.
3.3 Environmental variables
Most environmental variables were collected on the same date that I sampled the insects. Geographic coordinates in RT90 2,5 GON V were collected for each pond with a Garmin Dakota 20 handheld GPS with an accuracy of 5 meters and loaded with Friluftskartan™ Pro v3. After the insect sampling I measured maximum depth of each pond with a carpenter’s rule by wading out in the deepest part of the pond. A water sample for chemical analysis was collected with a 250ml plastic bottle from the middle of the pond, approximately 30 cm beneath the water surface. pH was estimated on site with a portable EcoSense® pH10 pH/temperature pen submerged in the water sampled for chemical analysis.
During the visits and sampling at each pond I also noted presence of fish, Great Crested Newt (Triturus cristatus) and Common Newt (Lissotriton vulgaris) in the pond by means of visual observations and catches during the insect sampling.
For an overview of the recorded environmental variables refer to table 1.
3.4 Vegetation cover and shoreline
Between August 28th and August 30th 2013 I revisited the ponds in order to estimate vegetation characteristics. Vegetation cover of the ponds was estimated visually in measures of tenths, ranging from no vegetation at all (0/10) to full cover (10/10). In addition, vegetation cover was recorded into three separate categories; floating leaved vegetation, submerged vegetation and emergent vegetation. I also estimated the percentage of barren shoreline (hard surface without vegetation e.g. stones, gravel) and bush vegetated shoreline (vegetation of a height of 1 meter or above within 2 meters of the shoreline). This was done
by walking around the ponds measuring total steps and steps with any of the two shoreline types.
3.5 Chemical analysis of water samples
The water samples collected were analysed for total phosphorus, total nitrogen and total organic carbon.
Total phosphorus was analysed using the method described by Menzel & Corwin (1965). In brief, organically bound phosphorus was transferred to
orthophosphate through oxidative hydrolysis with potassium persulfate and thereafter hydrolysis was performed in a vaguely acidic environment at high pressure and temperature using an autoclave. Afterwards the dissolved phosphate was measured using the Molybdate Reactive Phosphorus method where a spectrophotometer was used to measure the amount of
phosphomolybdenum complex to which the amount of phosphorus is proportional.
Total nitrogen was measured using the method described by Rand et al. (1976) in which all nitrogen in the sample was transformed to nitrate in the presence of a strong oxidizing agent. The nitrate was then analysed using a
spectrophotometer.
Total organic carbon was analysed using a Shimadzu TOC-‐L carbon analyser in which the sample is first freed from inorganic carbon, and oxidized under high temperature after which the resulting CO2 is measured with a non-‐dispersive IR gas analyser (Shimadzu brochure, Sugimura & Suzuki 1988).
3.6 Land use and GIS analysis
Terrain and land use around the ponds was estimated with the software ArcGIS 9 and the Terrängkartan™ map from Lantmäteriet. Land use was estimated in a 200-‐meter radius buffer zone along the shoreline of the ponds and excluded the pond area from the water surface land use category. Smaller ponds were not marked on the map and therefore I determined the centre of the pond (estimated from GPS coordinates) and drew a 200-‐meter radius circular buffer zone around the centre. The difference in total area between these two approaches differed by less than 5%, and therefore I concluded these measures to be comparable for estimating land use.
The following land use categories were estimated within the buffer zones:
coniferous and mixed forest, other open land, low-‐rise buildings, high-‐rise buildings, water surface, arable land, leisure homes, industrial land and precincts. In addition I estimated the distance from the ponds to nearest
developed area, and for each pond the distance to nearest inventoried pond. See appendix 3 for definitions of each land use category.
3.7 Statistical analysis
I explored and explained insect community structure and environmental variables using a multivariate ordination analysis. Since some of the
environmental variables are likely to be correlated I used a Principal Component Analysis (PCA) to reduce the 26 recorded environmental variables to
uncorrelated principal components (PCs). For explanation of the PC axes I considered the environmental variables with a factor loading of at least ±0,7 to be highly explanatory and included variables with a loading of at least ±0,5 for interpretation (Goertzen & Suhling 2013). These PCs were interpreted as meaningful ecological variables depending on their loading on each respective environmental variable. The PCs are listed depending on their explanatory value with PC1 axis projecting most of the variance, PC2 explaining the second most variance uncorrelated with the previous axis etc.
The relationship between insect community composition and environmental variables (the principal components) was explored by a constrained gradient analysis, redundancy analysis (RDA) using the statistical software R (R Core Team 2013). The insect abundance i.e. the total number of specimens per species and pond were used as variables.
In order to assess the insect diversity of the ponds I used the following metrics; i) species richness, ii) Shannon-‐Wiener diversity index and iii) insect abundance (total number of specimens found). I correlated these measures with the PCs using a backwards stepwise multiple regression to evaluate whether or not they were influenced by the PCs.
4. Results
4.1 Principal components on environmental variables
Nine PCs had eigenvalues ≥ 1, and together they explained 82,8% of the total variance (table 1). PC1 explained 19,4% of the variance and had a high loading on pond circumference, pond area and distance to closest pond as well as high pH but had high negative loading on nitrogen, phosphorous and carbon. PC1 is therefore associated with pond size and primary production. PC2 had high
loadings on surrounding forest, submerged vegetation, bushy shoreline and newt presence and a negative loading on open area suggesting that PC2 is
interpretable as woodland and vegetation. The high loading on distance to built area for PC3 imply that it represents remoteness from urban areas. It also has a high loading for emergent vegetation. PC4 had a high loading on adjacent waters (water surface) and is therefore associated with nearby water (excluding the ponds themselves). It also got a negative loading on adjacent low-‐rise buildings, roughly proposing a contrasting obstructive effect to water surfaces as semi-‐
suitable habitat corridors maybe facilitating connection between ponds.
Interpretation of the remaining 5 PCs are less straightforward since none of them have positive or negative factor loadings equal to or above 0,7 and in addition they explain less than 7 % of the variance each. PC5s highest factor
loading was floating leaved vegetation and represent the different species of vegetation with leaves on the water surface. PC6s highest loadings were on arable land and emergent vegetation. PC7 and PC8 had no variables that loaded more than 0,5. The relatively high loading of industry land on PC9 should be interpreted with care because there was a low absolute abundance of industry area in the dataset.
Table 1. Results of PCA on environmental variables at ponds.
Environmental Variables Factor Loadings and Interpretation
PC1 Size &
primary product-‐
ion
PC2 Forestat-‐
ion and vegetati-‐
on
PC3 Remo-‐
teness PC4 Other water areas
PC5 Floating vegetation
PC6 Field vegetati-‐
on
PC7
PC8
PC9
Industry
Pond Circumference 0,797 -‐0,084 -‐0,369 -‐0,072 0,165 -‐0,029 0,174 0,166 -‐0,006
Pond Area 0,779 0,105 -‐0,290 0,052 0,114 -‐0,048 0,031 0,321 0,171
pH 0,744 0,038 0,378 0,358 -‐0,094 0,153 -‐0,077 -‐0,105 -‐0,032
Total Nitrogen -‐0,713 -‐0,376 -‐0,194 0,001 -‐0,339 -‐0,039 0,220 0,116 0,032
Forest -‐0,082 0,791 0,036 0,348 0,021 -‐0,099 0,355 0,018 0,005
Other open land 0,368 -‐0,721 0,426 0,017 0,129 -‐0,156 0,232 0,052 -‐0,097
Distance to Built Area 0,222 -‐0,158 0,736 0,116 -‐0,169 -‐0,167 0,379 0,134 0,016
Floating Vegetation -‐0,677 0,044 -‐0,042 0,125 0,539 0,148 0,069 -‐0,093 -‐0,200
Distance to Closest Pond 0,647 0,045 -‐0,337 -‐0,240 0,222 -‐0,283 0,211 0,144 -‐0,118
Total Organic Carbon -‐0,632 0,016 -‐0,427 0,044 -‐0,358 -‐0,201 0,278 0,279 0,003
Total Phosphorous -‐0,533 -‐0,287 0,223 0,039 0,007 0,025 0,391 0,080 0,407
Common Newt presence -‐0,065 0,696 0,447 0,088 0,095 -‐0,138 -‐0,174 0,148 -‐0,131
Submerged Vegetation 0,134 0,671 0,192 -‐0,088 0,152 -‐0,091 0,089 -‐0,235 0,313
Bushy Shoreline -‐0,151 0,556 -‐0,397 0,441 -‐0,033 -‐0,029 0,326 0,091 0,145
Low-‐rise buildings -‐0,135 0,537 -‐0,387 -‐0,583 -‐0,246 0,072 -‐0,215 -‐0,070 -‐0,094
Crested Newt presence -‐0,049 0,524 0,414 -‐0,456 0,074 -‐0,102 0,046 -‐0,135 0,326
Emergent Vegetation 0,275 -‐0,003 0,506 -‐0,083 -‐0,290 0,532 -‐0,021 0,465 -‐0,031
Water surface 0,073 0,143 0,200 0,571 -‐0,151 -‐0,490 -‐0,347 -‐0,123 0,030
Arable land 0,184 -‐0,037 -‐0,051 0,122 -‐0,368 0,610 0,183 -‐0,403 -‐0,039
Industry -‐0,002 -‐0,407 -‐0,197 -‐0,040 0,461 0,169 -‐0,176 0,168 0,561
Depth 0,279 0,272 -‐0,338 0,040 0,293 0,183 0,397 0,040 -‐0,342
Fish Presence 0,440 -‐0,126 -‐0,407 0,276 -‐0,075 0,209 0,119 -‐0,481 0,248
Total vegetation -‐0,399 0,396 0,414 0,017 0,405 0,468 0,100 0,109 -‐0,051
Barren Shoreline -‐0,197 -‐0,447 0,264 -‐0,214 0,296 -‐0,254 0,274 -‐0,429 -‐0,188
High-‐rise buildings -‐0,386 -‐0,248 -‐0,206 0,489 0,312 0,131 -‐0,344 0,136 -‐0,074
Eigenvalue Magnitude 4,856 3,919 3,089 1,818 1,672 1,578 1,442 1,267 1,051
Variance Proportion 0,194 0,157 0,124 0,073 0,067 0,063 0,058 0,051 0,042
Bold: High factor loading ≥ 0,7; bold and italic: moderate factor loading ≥ 0,5.
4.2 Insect records, abundance and diversity
I recorded 65 autochthonous species; 18 species of Trichoptera with 3 groups of small larvae only determined to genus, 7 species of Odonata with 4 groups of small larvae only determined to genus, 12 species of Hemiptera with 4 groups of small larvae only determined to family level and 28 species of Coleoptera with 7 groups of small larvae determined to genus level. 29 of the total species only occurred in one pond (see table in Appendix 1). Species richness for the ponds varied between 1 and 20 species. The mean number of species per pond was 9,64 ± 6,11. The most commonly occurring species within respective order were:
Trichoptera; Limnephilus flavicornis (11 ponds), Odonata; Coenagrion puella/pulchellum (11 ponds), Hemiptera; Notonecta glauca (5 ponds), Coleoptera; Haliplus ruficollis (12 ponds) and Hygrotus inaequalis (8 ponds).
4.3 Insect assemblages and pond characteristics
The 9 RDA axes explained 43,7% of the total variance. The RDA found that species composition was significantly affected by PC3 (Remoteness) (p = 0,001) and PC6 (Field vegetation) (p = 0,035) and almost significant on PC2
(Forestation and vegetation) (p = 0,058) and PC4 (Other water areas) (p = 0,052). Hence the aquatic insect community was associated with remoteness to buildings, emergent vegetation (PC3), degree of surrounding arable land (PC6) and to some extent degree of surrounding forest and pond vegetation (PC2), as well as degree of low buildings and nearby water surface (Fig 2 & Fig 3).
A closer look at the RDA 1 and 2 showed that the response to any of the environmental variables is not uniform among any of the insect orders, but separate taxa respond differently. A few dragonfly species such as the
Coenagrion genus, Libellula quadrimaculata, Leucorrhinia pectoralis and Aeshna grandis seem to correspond well to PC1 and PC2 indicating the importance of pond size, nutrient levels, pond connectedness, vegetation and forest
surroundings. Lestes sponsa and undefined Aeshna and Lestes species seem to be affected by PC4, surrounding water surface and low-‐rise buildings (Appendix 2, Fig. A3).
Most Coleoptera species are clustered in the centre of the biplot indicating no or very weak responses to our tested environmental variables but some are
scattered, responding to certain PCs. Haliplus ruficollis is associated with PC4, Rhantus sp. with PC3, Hygrotus inaequalis with PC1 and PC2, Hyphydrus ovatus with PC8 and Haliplus immaculatus with PC7 (Appendix 2, Fig. A1).
Among the Hemiptera there were only a few examples of insects showing responses to the environmental variables, those included; Notonecta glauca which was weakly associated with PC2, and Ilyocaris cimicoides which was weakly associated with PC4 (Appendix 2, Fig. A2).
Many of the Trichoptera did not seem to associate with any of the PCs.
Limnephilus centralis might have a weak association to PC1, pond size and nutrient status (Appendix 2, Fig. A4).
Fig. 2. RDA ordination biplot of insect species and PCs, zoomed in to increase resolution. See
appendix 2 for biplots for specific insect orders. Insect species are abbreviated with first letter of genus and first three letters of species name. Groups are abbreviated with first three letters of genus followed by sp. For interpretation of PCs, see table 1.
Fig. 3. RDA ordination biplot of insect species and PCs. See appendix 2 for biplots for specific
insect orders. Insect species are abbreviated with first letter of genus and first three letters of species name. Groups are abbreviated with first three letters of genus followed by sp. For interpretation of PCs, see table 1.
4.4 Richness, abundance and diversity
Species richness was affected by PC3, PC7 and PC8. Hence more species rich ponds were further away from developed areas (Fig. 4) and had more emergent vegetation (Fig. 5) as well as a low influence from PC7 and PC8, which is
variation not explained by our tested environment variables. Abundance was affected by PC2 and PC3 and thus vegetation surrounding, bordering and in the pond positively affected the total abundance of insects. Presence of both species of newts is also positively affecting the total number of insects. There was a significant increase in both abundance of insects (t(25)abun= 3,25 p= 0,003) and species richness (t(25)sr= 2,22 p= 0,035) between ponds with no newts (Mabun= 28,9 SD= ±35,4; Msr= 7,3 SD= ±5,6) and ponds with one or both species of newts (Mabun= 116,2 SD= ±98,2; Msr= 12,4 SD= ±5,8). Adjacent low-‐rise buildings are also included into PC2 and have a small positive effect on insect abundance but increasing distance to built area is also positively correlated with abundance and the two may seemingly stand in contradiction to each other. Diversity was
affected by PC5 which is correlated to the amount of floating vegetation in the ponds including water lilies and duckweed (Table 2).
Table 2. Results of backward stepwise multiple regression.
Dimension Variables Unstandardized Coefficients Standardized Coefficients p-‐value Adjusted R2
Beta Standard Error Beta t
Richness Constant 9,269 0,683 0,000 13,570 0,000 0,674
PC1 0,547 0,310 0,202 1,765 0,093
PC2 0,664 0,345 0,220 1,925 0,069
PC3 2,055 0,389 0,604 5,287 0,000***
PC7 1,472 0,569 0,296 2,588 0,018*
PC8 2,359 0,607 0,444 3,889 0,001***
Abundance Constant 62,500 11,978 0,000 5,218 0,000 0,388
PC1 8,897 5,436 0,256 1,637 0,117
PC2 13,003 6,050 0,336 2,149 0,043*
PC3 21,849 6,815 0,501 3,206 0,004**
PC7 -‐15,110 9,975 -‐0,237 -‐1,515 0,145
Shannon Index Constant 1,466 0,107 0,000 13,709 0,000 0,385
PC2 0,094 0,054 0,272 1,733 0,099
PC3 0,095 0,061 0,244 1,557 0,136
PC4 0,121 0,079 0,240 1,529 0,143
PC5 0,207 0,083 0,392 2,500 0,022*
PC6 0,167 0,085 0,307 1,959 0,065
PC7 0,173 0,089 0,305 1,941 0,067
Fig. 4. Species richness plotted against distance to developed area. Species richness of aquatic insects increases with distance from ponds to developed areas.
Fig. 5. Species richness plotted against emergent pond vegetation. Species richness of aquatic
insects increases with the relative amount of emergent vegetation.
5. Discussion
5.1 The effect of urbanization on aquatic insect community structure
Distance from ponds to buildings and the amount of emergent vegetation where the two environmental variables that explained the majority of the variation in insect community structure in urban ponds. It emphasizes the importance of naturalness of the pond for many insect species and shows that urbanization can divide the insect community by supporting urban tolerant species and pushing away other species. This may imply that the urban landscape is negatively affecting the naturalness of the ponds and may actually reduce the pond quality.
Hence, the further away from developed area, the more suitable habitat are available for many aquatic insects. Booth & Jackson (1997) showed that urbanization degrades aquatic systems and that areas with just 10% of impervious soils (hard impermeable surface layers more common in urban
R² = 0,31132
0 5 10 15 20 25
0,0 50,0 100,0 150,0 200,0 250,0 300,0 350,0 400,0
Species Richness (n)
Distance to developed area (m)
R² = 0,37385
0 5 10 15 20 25
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8
Species Richness (n)
Emergent vegetation (frac)