Importance of Flight for habitat occupancy A pilot study on American Ground

Alexander Irwe

Degree project for Bachelor of Science in Biology

BIO602 Biology: Degree project 15 hec Spring 2015

Department of Biological and Environmental Sciences University of Gothenburg

Examiner: Christer Erséus Department of Biological and Environmental Sciences University of Gothenburg

Supervisor: Urban Olsson Department of Biological and Environmental Sciences University of Gothenburg

Frontpage photo by: Jeffrey S. Pippen Macon Co. NC12 May 2006

Table of content

Abstract…………………………………………………………………1

Introduction……………………………………………………………..2

Material and Methods…………………………………………………3

Results………………………………………………………………….5

Discussion………………………………………………………………5

Table 1 and 2…………………………………………………………..6

Acknowledgements……………………………………………………8

References……………………………………………………………..9

Abstract

Insects are one of the oldest groups of terrestrial still living today and were the first to evolve wings. Since then have evolved into many different shapes and sizes and occupy a wide range of niches. One group of insects, the Carabidae or ground beetles are interesting as they contain both with good flight capacity and species with reduced wings that have lost the capacity to fly. The genus is a group of ground beetles that have a larger than average flight capacity of most beetles. on the other hand is a genus were most of the species have reduced wings and lack the ability to fly. There is both cost and benefit in having flight capabilities and one benefit would be quicker and better movement. The assumption that flying species would better disperse themselves could be tested by comparing fundamental niche usage between two groups that differ in flight capabilities. To test this occurrence data of species from the Carabus and Cicindela genera were retrieved to calculate their realized niche, and Ecological niche modelling (ENM) was done to project their fundamental niche. The difference in realized and fundamental niche area was calculated for both genera and statistical tests were performed. The results indicate that Cicindela occupies a larger proportion of their fundamental niche than Carabus but the results from the statistical tests were not conclusive. A significant difference between the two genera was only seen when looking at the proportional difference but not when looking at the difference in real values. More species and further refinement of the ENM could help to increase confidence of the test.

Abstrakt

Insekter är en av de äldsta landlevande djur som fortfarande lever idag och var de första som utvecklade vingar. Insekter har sedan dess utvecklats till många olika former och storlekar och ockuperar en mängd olika nischer. En grupp av insekter, Carabidae eller jordlöpare är en intre ssant grupp då de innehåller både arter med god flygförmåga och arter med reducerade vingar som har förlorat sin förmåga att flyga. Släktet Cicindela är en grupp jordlöpare som har en flygförmåga över det vanliga för de flesta skalbaggar. Carabus är istället ett släkte där de flesta arter har reducerade vingar och saknar flygförmåga. Det finns både för och nackdelar av att kunna flyga och en fördel är att kunna förflytta sig snabbare och lättare. Antagandet att flygande arter skulle ha lättare att sprida sig kan testas genom att jämföra fundamental nischanvändning mellan två grupper med olika flygförmåga. För att testa detta hämtades data på förekomst av arter från släkterna Carabus och Cicindela för att sedan räkna ut deras realiserade nisch. ”Ecological niche modelling” (ENM) användes för att projicera arternas fundamentala nisch. Skillnaden mellan realiserad och fundamental nisch area räknades ut och stat istiska tester utfördes. Resultaten visar på att Cicindela ockuperade en större andel av sin fundamentala nisch än Carabus men resultaten från de statistiska testen var inte fullt övertygande. En signifikant skillnad mellan de två släkterna sågs endast för den proportionella skillnaden men inte för den faktiska skillnaden. Fler arter och mer finputsning av ENM kunde hjälpa till att öka säkerheten av undersökningen. 1

Introduction in almost all types of environments and can be found all around the globe. Most

of them are carnivores (Lövei et al, Insects were one of the first animals to 1996) but some species seem to be invade land some 400 million years ago omnivores or even frugivores (Hill et al, which created many new opportunities 1992). A lot of species have lost their for insects to adapt to new niches ability to fly or have reduced wings (Grimaldi et al, 2005). Since then insects (Lövei et al, 1996) but many species of have adapted to a number of different the genus Cicindela or common tiger environments and evolved many types which is part of the Carabidae of features. The ability to fly was first family can still fly (Carter, 1989). Another evolved by insects sometime in the large group of ground beetles is the Carboniferous time period and has since genus Carabus in which most species then evolved in many different species lack functional wings (Lindroth, 1986). and groups of animals (Knecht et al, Most species seem to 2011). Flight allowed for rapid movement have similar habitat and food preference to catch pray, avoid being caught or to (Lövei et al, 1996), so the major reach new otherwise inaccessible difference between the Carabus and habitats. Flying is costly since it requires Cicindela genera are the flight a lot of energy to remain airborne, so for capabilities. it to evolve the benefit has to outweigh the cost of maintaining the ability to fly One can then assume that species from (Roff, 1986). When studying newly the Cicindela group would be much formed habitats such as islands made more proficient in finding new areas with from volcanic activity, species with flying suitable habitats. Since most of the capabilities and plants with air-travelling Cicindela species still have functional seeds and pollen are usually the first wings it would be assumable that they ones to settle in the newly formed have an easier way to disperse habitat. Interestingly enough, islands themselves. One possible way of testing and other isolated habitats have a this would be to calculate the area of the proportionally larger number of species realized niche for the species in each without flight capabilities than inland group and then calculate the area of the habitats. This suggests that selection fundamental niche for each species. pressure for flight may decrease in Realized niche in this case refers to a isolated islands (Gillespie, 2007). The species current niche where it actually term brachypterous is used to describe resides and fundamental niche refers to the loss of or reduced wings in animals a species potential niche were it could and can be seen throughout the Insecta live (Sadava et al, 2011). A species lineage. Some species have both realized niche is usually smaller than its winged and brachypterous morphs in a fundamental niche as it might be population and are thought to be an restricted from a number of factors such adaptation depending on the stability of as competition from other species or the environment or a trade-off from other migration hinders. If Cicindela species features (Roff, 1986). One obvious are better at dispersing they would use drawback from wing reduction is the loss their potential habitats to a higher extent in mobility and perhaps the potential than species from the Carabus genus. In ability to disperse. this study I gathered information on the

current distribution and compared it to One interesting group of insects are the the potential distribution estimated by ground beetles Carabidae which today Ecological niche modeling. The test is to have more than 40,000 species see if flight capability has a positive described (Erwin, 1985). They are found 2

effect on the ability to disperse. The null points together with environmental data hypothesis is that no difference in to model the species fundamental niche relative utilization of available habitat over a geographical area (Kearney, exists between two groups of different 2006). Before the initial ENM was done flight capabilities. A difference would a trial and error phase was conducted to violate the assumptions of the null see which environmental layers were to hypothesis and indicate that flight is be used in the final model. This was important for the ability to colonize all done using a custom made workflow areas of suitable habitat. based on jackknife resampling that test which of the chosen layers affected the Material and Methods model the most (Leidenberger et al). The layers that had the biggest effect were Occurrence data retrieval and considered most important for the Ecological niche modelling accuracy of the model. Correlation tests using ENMtools For this study 10 species of each ground (http://enmtools.blogspot.se/) were also beetle genus Carabus and Cicindela conducted to make sure none of the were used, giving a total of 20 species. layers correlated with each other. If any More species would have been selected layers correlated the one that had the but there was a limited number of lowest effect from the jackknife tests was Carabus species with occurrence data removed. The environmental layers available. North America was selected selected by the jackknife and correlation as an area of interest since it has a lot of tests were: annual mean temperature, well documented records of species and precipitation of driest month, is a relatively large area with no large temperature annual range, bulk density, expanses of water interfering with salinity, gravel content, ph and altitude in species migration. The selected species meters. Studies on ground beetles were chosen by looking at which of them suggested that humidity would be a had the most available data from the limiting factor for habitat preference Global Biodiversity Information Facility (Lövei et al 1996) which coincides with (GBIF) (http://www.gbif.org/). The the precipitation of driest month layer in occurrence data for each individual the model. Sand fraction was also added species was retrieved using the Data as a layer because a study on habitat Refinement Workflow v17 available at preference for eight Cicindela species the Biovel portal showed a preference for sandier ground (https://portal.biovel.eu/). This workflow types (Schultz, 1989). The occurrence is connected to GBIF for automatic data from all the species were also used species occurrence retrieval. After the in the Bioclim workflow (also available at occurrence data was retrieved a the Biovel portal) to only select data refinement of the data was done using points that were environmentally unique Open Refine (http://openrefine.org/) to be used in the ENM. The same which is a part of the Data Refinement selected layers were both used for the Workflow. This refinement included Bioclim workflow and the ENM. The removing any data points lacking valid models were created using the default coordinates, any data points that were Ecological niche modelling workflow at older than 1970, and excluding any data the Biovel portal. The ENM algorithm points not found in the United States of was set to Maximum Entropy and the America or Canada. To calculate the default algorithm parameters were used. area for the species potential habitats I The selected layers were projected in used Ecological niche modelling (ENM). 10arc-minutes and cross validation on This process uses occurrence data 10 replicates, with threshold set to 3

lowest presence point, was done in Equidistant Cylindrical coordinate order to test the accuracy of the models. reference system that uses meters as The projections made from the models length unit. Each projection has a set of were not projected on the entire North values indicating the habitat suitability American continent as some areas were for a particular species were the highest left out using a mask created for the value equals 100% suitability and the projections. This was done by mapping lowest 0% suitability (Fig. 2). A new all findings of beetles in North America raster tif-file was generated using the by looking at a map from GBIF showing raster calculator of QGIS. The new all beetle findings (Fig. 1). Projections raster file only included values higher from all the species was saved as tif-files than 40-50% suitability depending on the for further analyses. strength of the projection model. This new raster file was then transformed into a vector file for easier area calculation using the area function included in attribute tables in vector files. To calculate the realized niche area of a species, a buffer zone was made around each occurrence point forming another vector file. The buffer zones were approximately 150 km in diameter and the summed area for all buffer zones were also calculated using the area function in the attribute tables. All of the Figure 1. Map showing occurrence data points above steps were done for each of the from all species of beetles in North America. 20 species.

Area calculation using QGIS Statistics using SPSS

To calculate the specie’s realized and To test the significant difference fundamental niche area, the free open between the two groups a statistical test source program QGIS was used was performed using the software IBM (http://www.qgis.org/en/site/). The SPSS Statistics 22. Before the actual occurrence data points and the test a minor analysis was performed to projection tif-files were imported into see if the data was normally distributed QGIS and projected using the World as this is a requirement for certain

Figure 2. The projection tif-file for Carabus chamissonis showing the suitability and occurrence points.

4

statistical tests. The difference in area map of the realized and fundamental between the realized and fundamental niche area for one of the species after all niche of the two groups was tested by a the steps in QGIS were made. Cicindela simple independent T-test with a has a higher average realized niche confidence interval of 95%. This was area of 1006587 km2 and Carabus an done for the actual values but also on average value of 372296 km2. Difference the ratio between realized and in fundamental niche area is not as high fundamental area divided by the between the two genera as both are fundamental one. This is done since the close to 2000000 km2. The difference realized niche area differs between between the realized and fundamental some of the species and having it in niche is slightly higher in Carabus with percentage gives a better representation an average value of 1589605.4 km2 as of the difference. The percentage values opposed to Cicindela with an average of were also arcsin transformed to make 1207211.3 km2 (Table 1). The statistical them closer to a normal distribution but test does not show a significant also because values near 100 or 0% difference between the groups for the could hamper with the statistical test. actual values. The transformed values (Chestnut, 1995). do show a significant difference between the groups (Table 2). Results Discussion The number of data points differed between the two genera as the Cicindela When looking at the actual results from group had a larger average of 93 data Table 1 there is indication of a difference points and Carabus only had an average between the two genera. Carabus has a of 31 data points per species (Table 1). lower average realized niche area and The average internal area under the there is a bigger difference between curve (AUC) from the ENM for all the realized and fundamental niche area for models was 0.955. The AUC is a sort of Carabus than for Cicindela. This could measure of model strength and a value indicate that the assumption made above 0.8 for this study is considered earlier is correct. Species with lower acceptable (Hanley et al, 1982). flight capabilities are worse at using their Average omission error for all species potential habitat range than species with was 8.03 which are at an acceptable good fight capabilities. level for this study. Figure 3 shows a

Figure 3. A map of the realized and fundamental niche area of Carabus chamissonis after all QGIS steps.

5

Table1. Showing data points, realized and fundamental niche area, difference and arcsin

transformed proportional difference for all the species of the two genera.

Proportional Data Realized niche Fundamental difference Species points area (km2) niche area (km2) Difference (arcsin)

Cicindela formosa 62 664195 1880915 1216720 53.54

Cicindela hirticollis 62 876397 1914265 1037868 47.42

Cicindela limbalis 43 549719 2717214 2167495 63.27

Cicindela longilabris 41 589651 2241531 1651880 59.14

Cicindela oregona 78 956365 1415283 458918 34.71

Cicindela purpurea 53 844475 1464253 619778 40.59

Cicindela repanda 170 1093278 1916588 823310 40.95

Cicindela scutellaris 63 1017693 3333723 2316030 56.46

Cicindela sexguttata 260 1696390 1913007 216617 19.66

Cicindela tranquebarica 99 1777707 3341204 1563497 43.16

Carabus chamissonis 45 452777 2985853 2533076 67.08

Carabus granulatus 11 126462 2032449 1905987 75.56

Carabus maender 28 355631 2304385 1948754 66.87

Carabus nemoralis 41 554612 1572411 1017799 53.57

Carabus rossii 7 204464 1334504 1130040 66.96

Carabus sylvosus 5 151497 3585108 3433611 78.14

Carabus taedatus 137 1362728 1951764 589036 33.32

Carabus truncaticollis 12 180154 1464045 1283891 69.46

Carabus vietinghoffi 19 116203 534035 417832 62.19

Carabus vinctus 7 218432 1854460 1636028 69.93

Mean Cicindela 93 1006587 2213798,3 1207211,3 45.89

Mean Carabus 31 372296 1961901.4 1589605.4 64.31

Table 2. Results from the statistical t-test showing significance for the proportional values but not from the real.

Values Sig. (2-tailed) value Significant Real 0,310 No Arcsin proportional 0,005 Yes

6

The evidence for this is not clear though. the models. Competition and predation The results from the statistical t-test from other species are also not taken could both be interpreted as significant into account in the modelling. Studies or not. One can both argue for using the have shown that habitat preference for actual or transformed values from the species of Carabidae are among other statistical test. Some of the species, things influenced by number of especially from the Carabus genus had competitors in the area (Lövei et al, very few data points, thus giving a large 1996). To increase the accuracy of the difference between realized and models one could map out where the fundamental niche area. The numbers competitors have their habitat range and would differ greatly between species use that as a factor in the models. Wind resulting in outliers interfering with the current and strength could also perhaps test. This is somewhat fixed by using a in some way effect flying individuals percentage ratio instead reducing the disperse patterns. The reason for not intensity of outliers. The aim of this study projecting the models on the entire North was to see if there was any difference in American continent and only project on fundamental niche usage between the areas known to have been sampled from two genera so the actual value was to try and increase the accuracy of difference is not as important in this the models. If the unsampled areas case. The real interest is in the would have been included the results proportional difference of fundamental could have been biased since one niche usage. The real value difference cannot know if the unsampled areas do could be of some interest but should not contain the species of interest or not represent this study’s main result. Even thus projecting suitable area where it so, by having more species of each might actually reside. The calculations of genus could perhaps make the statistical the fundamental and realized niche test stronger showing significance from areas were also quite rough and could the actual values as well. As stated possibly be done with more accuracy. earlier Carabus was the limiting genus Time and resources was a limiting factor only having around 10 species with for this study so the calculations of the available data points at GBIF. Carabus areas were not as precise as hoped for. also had a lower average number of The buffer zones around the occurrence data points per species. This could lead points are a rough estimate of the to a difference in model accuracy beetle’s habitat as the area they live in between the two genera as seen in this around that point could very well be case, where Carabus had a larger more than 150 km wide. Also points average external omission error than located close to the shoreline or near the Cicindela when performing the cross edge of large water masses could validation. possibly include large areas of water where the ground beetles do not reside. The limited number of species was not For further development of this study all the only thing that could interfere with of these sources of error have to be the results. The Ecological niche thought about and one would have to try modelling available at the Biovel portal to eliminate them. has a limited number of environmental layers to choose from when creating the Studies on migration and species model but there might be other aspects dispersal could be key factors to help in that determine a species habitat range. the understanding of evolution and None-biological factors such as human speciation. One way of many for new agriculture or infrastructure could limit a species to arise is through migration of species range but are not considered in curtain members of a population that 7

later isolate themselves from the rest flying. This could also be applied on (Sadava et al, 2011). One way of other flying animals such as birds and migration is of course to fly to reach new bats. suitable areas. To study the impact of flight on migration and other aspects Conclusion could lead to better understanding of speciation. The research on why and There is indication of difference in when species evolve wings is another fundamental niche usage between the interesting field of study that with the two genera Carabus and Cicindela. The help of migration studies could increase genus Carabus containing species with our understanding of winged animals. reduced wings seems to have a harder Flight capability seems to have a time reaching new suitable habitats. positive effect on speciation as insects Cicindela is using its fundamental niche are the largest group of animals today to a fuller extent than Carabus. The with almost one million described statistical test shows a significant species and an estimate number of 2.5 difference depending on which values to 10 million species to exist (Grimaldi et are tested. The actual value difference al, 2005). Another great example of between realized and fundamental niche species richness in flying animals are of area shows no significance but when course the birds with around ten looking at the proportional difference thousand species alive today (Clements, there is a significant difference. To 2007). Bats are another group of increase the power of the statistical test, animals capable of flight and they as more species could help in making the well show a great species richness with difference significant even for the actual around a thousand species being the value difference. Further development of second biggest mammal group after the the Ecological niche models could also rodents (Tudge, 2000). The connection make the experiment more accurate. between flight capabilities and species This indication of flight importance for richness is an interesting subject and habitat occupancy will hopefully inspire increasing our understanding of more people to investigate in this subject migration of flying species will definitely and together we might increase our help in the research of such understanding of winged individuals. connections. Acknowledgements An interesting continuation of this study would be to test more species for the Matthias Obst Researcher and project two groups or to use different genera. leader Department of Biological and Carabus mostly have reduced wings Environmental Sciences, Systematics (Lindroth, 1986) and most of the and Biodiversity Göteborg University Cicindela species have good flight Helping with ENM. capabilities (Carter, 1989), so it would be interesting to see the results from a Sarah Bourlat Researcher genus that is something in between Department of Biological and Carabus and Cicindela that has Environmental Sciences intermediate flight capabilities. One step Systematics and Biodiversity further would be to look at different types Göteborg University of insects and see if they also differ in Helping with ENM. niche usage depending on flight capabilities. Different types of moths, Peter Tiselius Professor in marine butterflies, bees and wasps certainly zooplankton ecology Department of differ in their speed and efficiency when Biological and Environmental Sciences, 8

Systematics and Biodiversity Göteborg Hanley, James A. McNeil, Barbara J. University 1982. The meaning and use of the area Helping with statistics. under a receiver operating characteristic

(ROC) curve. Radiology Vol. 143: 29-36. Donald Blomqvist Researcher

Department of Biological and Hill, James M. Knisley, C Barry. 1992. Environmental Sciences Frugivory in the , Cicindela Systematics and Biodiversity repanda. The Coleopterists Bulletin Vol. Göteborg University 46: 306-310. Helping with statistics.

Kearney, M. 2006. Habitat, Environment Urban Olsson Associate professor and niche: what are we modelling?. Department of Biological and Oikos Vol. 115: 186-191. Environmental Sciences

Systematics and Biodiversity Knechta, Richard J. Engel, Michael S. Göteborg University Bennera, Jacob S. 2011. Late Supervisor during the project. Carboniferous paleoichnology reveals the oldest full-body impression of a flying References . PNAS Vol.108: 6515-6519.

Carter, Mark R. 1989. The Biology and Leidenberger, Sonja. Giovanni, Renato Ecology of the Tiger Beetles De. Kulawik, Robert. Williams, Alan R. (Coleoptera: Cicindelidae) of Nebraska. Bourlat, Sarah J. 2014. Mapping present Transactions of the Nebraska Academy and future potential distribution patterns of Sciences Vol. 17: 1-18. for a meso-grazer guild in the Baltic Sea. Journal of Biogeography. Chestnut, Dexter F. 1995. Analysis of statistical tests to compare visual analog Lindroth, Carl H. 1986. The Carabidae scale measurements among groups. Anesthesiology Vol. 82: 896-902. (Coleoptera) of Fennoscandia and Denmark. Vol. 2. Brill. Clements, James F. 2007. The Clements Checklist of Birds of the Lövei, Gábor L. Sunderland, Keith D. World. Comstock Publishing Associates. 1996. Ecology and Behavior of Ground Sixth Edition. Beetles (Coleoptera: Carabidae). Annual Reviews Entomol Vol. 41: 231-256. Erwin, Terry L. 1985. The taxon pulse: a general pattern of lineage radiation and Roff, Derek A. 1986. The Evolution of extinction among carabid beetles. Wing Dimorphism in Insects. Evolution Phytogeny and Zoogeography of Vol.40: 1009-1020. Beetles and Ants. Junk Publishers, Dordrecht. Sadava, David. Hills, David M. Heller, H Craig. Berenbaum, May R. 2011. Life Gillespie, Rosemary G. 2007. Oceanic The Science of Biology. Sinauer Islands: Models of Diversity. Associates. Sunderland USA. Encyclopedia of Biodiversity. University of California, Berkeley.

Grimaldi, David. Engel, Michael S. 2005. Evolution of the Insects. Cambridge University Press. 9

Schultz, T. D. 1989. Habitat Preferences and Seasonal Abundances of Eight Sympatric Species of Tiger Beetle, Genus Cicindela (Coleoptera: Cicindelidae), in Bastrop State Park, Texas. The Southwestern Naturalist Vol. 34: 468-477.

Tudge, Colin. 2000. The variety of Life Vol. 2 Oxford University Press.

10