Cytotype distributions of the common Dandelion (Taraxacum section Ruderalia) and the
Cuckoo flower (Cardamine pratensis) in Europe and how these are affected by climate
change
Bachelor project of L. Staal Supervisor: P. Meirmans
Department of Evolutionary and Population Biology, University of Amsterdam, Amsterdam, the Netherlands Abstract - Polyploidy is a common phenomenon in plants and can occur through all sorts of genetic processes. Some species of plant contain different cytotypes. These cytotypes may differ in fitness and habitat preferences. Two such species are the common Dandelion and the Cuckoo flower, both species consist of multiple cytotypes. Diploids of both species are mainly found in Central Europe while polyploids are found more in the north, up to Scandinavia. In this study the focus is on the different cytotype distributions and the effect climate change may have on these distributions. With the use of a meta-analysis and SDM, current and future distributions were modelled. Current models of the common Dandelion corresponded with the already known distributions. Both cytotypes seemed to move away from the most southern parts of their distribution if no actions were taken against climate change. Cuckoo flower data was too biased to make accurate distribution models.
Introduction
Polyploids are commonly found in land plants and have been documented several times in the evolutionary history of these plants (Castro et al., 2019). Following polyploidisation, the evolution of polyploid genomes includes all sorts of genetic processes such as recombination, gene silencing and chromosome rearrangement (Lihova et al., 2004). All these processes can affect the cell size and gene expression of the polyploids in comparison to their diploid progenitors, which may lead to differences in environmental and ecological preferences (Castro et al., 2019). Because of this, the different cytotypes often occupy different geographic ranges resulting in distinct distributions (Castro et al., 2019). In most species, the cytotypes 2x and 4x are dominant, meaning that these cytotypes have an overall frequency higher than 20% and are most often observed in single-cytotype populations. This higher frequency is likely due to higher fertility as a result of balanced chromosome pairing (Kolář et al., 2017). Most polyploid species also include odd-ploidy cytotypes, which can play a role as mediator of gene flow (Kolář et al., 2017). In several species, multiple cytotypes populate the same areas and establish mixed populations (Kolář et al., 2017).
Climate change and human induced disturbance, such as anthropogenic land-use, are thought to play prominent roles in the extinction and changes in distribution of many plant species (Thuiller et al., 2005). The changes in land-use are strongly driven by population growth and urbanisation, which increases the demand for food and other resources (Maes et al., 2015). Often the land to produce these resources and space for housing comes with the loss of wild lands, which play a huge part in the preservation of biodiversity (Maes et al., 2015). Climate change causes the earth’s surface to heat up, meaning that the average temperature on earth is increasing (Stocker et al., 2014). It is likely that the frequency of heat waves will increase in Europe, and that precipitation will be more frequent and heavier (Stocker et
al., 2014). For the Mediterranean region it is likely that intensity and frequency of droughts will increase (Stocker et al., 2014). Summers in Europe increase in temperature faster than the world mean (Stocker et al., 2014). Since different
cytotypes have different ecological preferences it is likely that these alterations will change the distribution of the different cytotypes.
The common Dandelion
One species containing multiple cytotypes is Taraxacum officinale (common Dandelion), which has cytotypes ranging from diploids to tetraploids (Verduijn et al., 2004a). Pentaploids and hexaploids can arise by reproduction of triploids and tetraploids but these cytotypes have rarely been recorded and are not well studied (Mártonfiová, 2006). The common Dandelion grows in grasslands, roadsides and lawns on wet to dry ground (van der Meijden, 2005, p.630). Diploid dandelions are mostly distributed in central Europe where their range overlaps with that of the triploids whose range stretches up to northern Scandinavia (Verduijn et al., 2004a). Although diploids and triploids are both abundant, tetraploids are rare, this is probably due to their low fertility (Verduijn et al.,2004a). There are three hypothesis for the different distribution of the cytotypes: 1) the different reproductive strategies used by the different cytotypes, 2) triploids have a higher tolerance against human disturbance and 3) diploids are more thermophilic compared to the triploids.
The difference in distribution may be connected to the difference in reproductive strategies of the diploid and triploid individuals (Verduijn et al., 2004a). Diploids reproduce sexually while triploids are obligate asexuals. Frequently, hermaphroditic triploids produce seeds asexually but still produce pollen. When these pollen reproduce with diploids their offspring will also be triploid individuals (Verduijn et al., 2004a). The fact that triploids can reproduce asexually means they have an advantage in populating new areas, this is a possible explanation as for why triploids are more abundant in the areas that were once glaciated (Asker & Jerling, 1992).
Meirmans et al. (1999) investigated which ecological differences could play a role in the distribution of the diploid and triploid dandelions. They studied Dandelions in Neuchâtel, Switzerland where sites differed in altitude and disturbance. The distribution of diploids and triploids was found to be mainly associated with human disturbances like mowing, but also ecological indicators such as humidity and vegetation type. One indicator they looked at for human disturbance was the coverage of therophytes. They found that both this and altitude were significantly correlated with the presence of the different cytotypes. Triploid abundance was positively correlated with therophyte coverage, while diploid abundance was positively correlated with altitude. It is known that coverage of therophytes indicates human disturbance. However, it is likely that high altitudes was also correlated with human disturbance, since in that region the population density, and hence the degree of human disturbance, is lower at higher altitudes. This could mean that a high amount of human disturbance favours the distribution of triploids over diploids
The northern distribution boundary of the diploids runs through Belgium and the Netherlands (Roetman et al., 1988). In the southernmost part of the Netherlands, Roetman et al. (1988) found diploid individuals intermixed with the more common triploid populations. They speculated that diploid individuals might have more thermophilic habitat preferences than triploid individuals. Elzinga et al. (1987) looked at the province Limburg specifically and also found
that diploids are slightly more thermophilic than triploids, but also speculated that diploids may occupy drier habitats compared to triploids. Verduijn et al. (2004b) later experimentally confirmed the thermophilic habitat preference of the Dutch diploids. Therefore it is expected that with a warming climate, the distribution of the diploids will move further northwards.
The Cuckoo flower
Another species containing multiple cytotypes is Cardamine pratensis (Cuckoo flower). This species grows on wet to moist ground that is rich in nutrients, like grasslands, deciduous forests and swamps (van der Meijden, 2005, p.423-424). The Cuckoo flower has several cytotypes ranging from diploid to dodecaploid (Berg, 1967). Because the basic chromosome number of the Cuckoo flower can be eight or seven, the variety of euploids can contain all different combinations of these numbers. This means that for example octoploid do not only have cytotype 2n=64 or 2n=56, but also 2n=58, 2n=60, 2n=62 and 2n=64. This results in cytotypes ranging from 2n=16 to 2n=96 (Berg, 1967). They are distributed throughout Europe, northern Africa, northern North America and Asia, though most research is done in Europe (Lihova et al., 2003). The most common cytotype in Europe is hypotetraploid (2n = 4x-2 = 30). Diploids (2n=16) of this species have been found in Central Europe, Belgium and France while higher polyploids (2n = 48 or 56) have been found in northern Europe (England, Finland, Denmark and Sweden) (Lihova et al., 2003). The Cuckoo flower reproduces sexually and are self-incompatible meaning they need to be cross pollinated (König et al., 2014). All cytotypes seem to be able to also use vegetative reproduction methods, although it seems to be more commonly used among the higherploids in the wild. This could be because higherploids are often associated with habitat conditions most favourable for vegetative reproduction (Salisbury, 1965).
Although many studies have focused on the cytotypes of these two species, no database of exact locations
representing the distributions of the different cytotypes exists. This study tries to create such a dataset using a meta-analysis. This dataset will be used to construct how the different cytotypes of the common Dandelion and the Cuckoo flower are distributed and how these distributions may change in the future due to climate change in Europe. Data will be obtained from the literature which will be used to construct a species distribution model for the separate ploidy levels. We will then use this model to forecast the distribution of the cytotypes under different projections of the future climate. It is expected to find most diploids for both species in central Europe, while triploids and polyploids will mostly reside in northern Europe. In the future, diploids of the common Dandelions are expected to expand their distribution more northwards as temperatures in whole Europe rise. Although not much is known about the preferences of the different cytotypes of the Cuckoo flower, the diploids are known to follow a similar general distribution as the common Dandelion diploids. Therefor a hypothesis may be that the diploids will also expand their distribution northwards.
Methods Meta-analysis
Data for this study was collected from published research papers. Data was included in the database for a species if it included the geographical location and chromosome number or ploidy level. Species names for the common
Dandelion must fall under the section Ruderalia. Since this section has had multiple names over the course of time the following three names will be included: Taraxacum sect. Taraxacum, Taracaxum sect. Vulgaria and Taraxacum sect. Ruderalia. For the Cuckoo flower, all subspecies in the Cardamine pratensis group were included. If a paper included geographical coordinates for a location, these were included in the database. However, most papers only contained a description or a name for the location. For these locations the coordinates were extracted using the Google Earth software (Google LLC., 2020). When only a city or a village name was given, the coordinates of the city centre were taken. When a more specific location was given, such as: “near the river”, “5 km north of” or “between” this information was used for obtaining the coordinates. If only a direction (i.e. north of) or the word “near” was used, the city centre was used since no specific location could be extracted. Only locations within Europe were included, to avoid including invasive populations. For the common Dandelion, only diploid and triploid presence was noted , since tetraploids were almost never mentioned. For the Cuckoo flower every cytotype was noted down, all cytotypes consisted of a number of chromosomes. An unpublished dataset, containing location data of the common Dandelion cytotypes, of Meirmans was added to the Dandelion data. Each data point for the common Dandelion consisted of the genus, section, location, country, longitude, latitude, reference and the presence or absence of diploids and triploids. The Cuckoo flower data consisted of species, subspecies, location, country, longitude, latitude, reference and the presence or absence of every known cytotype.
Bioclimatic variables
GIS layers of bioclimatic variables for future and current climate were obtained from Worldclim (Hijmans & Elith, 2012), with a resolution of 2.5 min. Both current and future sets contain 19 bioclimatic variables. The future dataset used was for the period 2081 – 2100 with data from the MIROC6 climatic model for four SSP’s: SSP 126, SSP 245, SSP
370 and SSP 585. SSP stands for shared socioeconomic pathway, such pathways represent different future climate scenario’s. It uses climatic data and socioeconomic factors to create models for different climate policy scenario’s (Explainer, 2018). SSP 126 is the best case scenario, while SSP 585 is the worst case, SSP 245 and SSP 370 are two scenario’s in between. To avoid multicollinearity, all variables were tested for correlation; when the correlation coefficient was 0.7 or higher, one of the variables was excluded from the data set. The choice of which variable to exclude depended on which one was easiest to interpret. As a result, 7 bioclimatic variables were retained: Annual mean temperature (bio 1), Mean Diurnal Range (bio 2), Temperature Seasonality (bio 4), Precipitation of Wettest Month (bio 13), Precipitation of Driest Month (bio 14), Precipitation Seasonality (bio 15) and Precipitation of Coldest Quarter (bio 19).
Data preparation
First the location data of the two flowers was loaded into R. Since locations without coordinates cannot be used in the analysis, all locations where no coordinates were found using the locations from the papers were removed. Most often this was due to the location not being known to Google Earth or multiple location having the same name, without being able to distinguish which place corresponded with the one in the paper. The common Dandelion dataset was then split in two, one representing the locations where diploids were found, one representing the locations where triploids were found. The Cuckoo flower data was first divided into ploidy groups, these groups were small and therefor needed to be divided into subgroups containing multiple ploidy levels. Each ploidy level of the Cuckoo flower was plotted, the ploidy levels with similar distribution patterns were put together in a group to use for the species distribution models. This resulted in three groups: diploids, mediumploids (triploids, tetraploids,
pentaploids and hexaploids) and higherploids (septaploids and higher ploidy levels). After this the data for both the common Dandelion and Cuckoo flower subgroup was cleaned, starting with removing all duplicated locations from the subgroups. This was followed by checking similarity between country label and coordinates, if coordinates did not match the country mentioned the data point was removed. To remove the sample bias, the subgroups were put in a raster with 2 degrees grids, one sample location was selected per grid so the sample did not overrepresent the most sampled areas. For every subgroups 5 subsamples were made using this method and were used in the species distribution models.
Species distribution models
Species distribution models (SDM’s) are a frequently used tool to predict the distribution of a species. These models use presence/absence data and environmental data to look for correlations and with that look for potential suitable habitats. These types of models are also often used to see how species distributions might be effected by climate change (Elith & Leathwick, 2009).
Four distribution models were made for every subsample, a bioclim model (Hijmans et al., 2017), a general linear model (GLM) (Stats Package | R Documentation, n.d.), a random forest model (Cutler & Wiener, 2018) and a support vector machine model (SVM) (Karatzoglou et al., 2019). Modelling was performed using the seven selected
bioclimatic variables. For the species distribution models for future scenarios the same seven variables were used but from the future data sets. For all ploidy group the subsample models were combined to get an average per model. This was followed by combining all four models of all 5 subsamples for each ploidy group. All combined models were weighted using the AUC of the individual models. For each ploidy group, the results of every model and all
subsamples were combined into an average overview of the distribution. By combining these models the AUC of the individuals was used again for the weighted mean.
For both data preparation and species distribution modelling Hijman and Elith’s (2017) tutorial for Species
distribution modelling with R was followed. The method described in the tutorial for removing the spatial sorting bias (ssb) was not used since the sample bias removal mentioned in data preparation was deemed sufficient enough for removing any ssb.
The data repository, containing all datasets and R-scripts, is stored on zenodo.org, under the name Data repository: Cytotype distributions of the common Dandelion (Taraxacum section Ruderalia) and the Cuckoo flower (Cardamine
pratensis) in Europe and how these are affected by climate change (DOI: 10.5281/zenodo.3944635). Results
Meta-analysis
Location data was obtained from 17 papers, nine for the common Dandelion (in addition to the unpublished data of Meirmans) and eight for the Cuckoo flower (see table 1). This resulted in 692 data points representing the common Dandelion and 428 representing the Cuckoo flower. The dandelion data contained 240 diploids and 622 triploids. The Cuckoo flower data contained 41 different cytotypes, which are noted down in appendix 1 along with their presence
numbers. The countries that were included in the papers were: Austria (D: n=38, C: n=37), Belgium (D: n=138), Bulgaria (D: n=1), Croatia (D: n=2), Czech Republic (D: n=3, C: n=6), Denmark (D: n=1), France (D: n=71, C: n=12), Germany (D: n=122, C: n=30), Greece (D: n=1), Hungary (D: n=17, C: n=7), Italy (D: n=2, C: n=10), Liechtenstein (C: n=2), Montenegro (D: n=1), Netherlands (D: n=156, C: n=11), Poland (D: n=59, C: n=8), Portugal (C: n=5), Romania (D: n=5, C: n=14), Scotland (D: n=3), Serbia (D: n=1), Slovakia (D: n=11, C: n=83), Slovenia (D: n=1, C: n=7), Spain (C: n=32), Switzerland (D: n=48, C: n=93), UK (C: n=19) and Ukraine (C: n=51).
Table 1. An overview of which references are used in the meta-analysis and which cytotypes were present in each
reference.
Common Dandelion Cuckoo flower
reference cytotypes reference cytotypes
Calame & Felber, 2000 Diploids, triploids Berg, 1967 28, 30, 31, 32, 34, 36, 38, 39, 45, 46, 52, 53, 54, 56, 58, 59, 60, 62, 64, 66, 67, 68, 69, 70, 72, 74, 76, 78, 84, 118 Elzinga et al., 1987 Diploids, triploids
Jenniskens et al., 1984 Diploids, triploids Dale & Elkington, 1974 56, 58
Meirmans et al., 1999 Diploids, triploids Lihová et al., 2003 16, 24, 30, 32, 40, 44, 46, 48, 56 Meirmans, unpublished Diploids, triploids Marhold, 1989 16, 18
den Nijs & Sterk, 1980 Diploids, triploids Marhold, 1994 16, 17, 18, 19, 20, 21, 24, 30, 32, 38, 44, 48, 80
den Nijs & Sterk, 1984a Diploids, triploids Marhold, 1996 16, 30, 38, 44 den Nijs & Sterk, 1984b Diploids, triploids Marhold, 2000 16, 18, 32 den Nijs et al., 1990 Diploids, triploids Urbanska-Worytkiewicz
& Landolt, 1974
16, 17, 18, 19, 21, 24, 30, 32, 38, 40, 46, 48, 54
Roetman et al., 1988 Diploids, triploids
Bioclimatic variables
Table 2 shows the correlation matrix between all 19 bioclimatic variables. Variables 2 and 15 were not correlated to any other variables. Variable 1 excluded variables 3, 5, 6, 8, 9,10 and 11. Variable 4 was correlated to variable 7, of these two, variable 4 was selected. Variables 13, 14 and 19 were chosen due to few correlations with other variables.
Table 2. Correlation coefficient matrix of the 19 bioclimatic variables from Worldclim. Correlation coefficient ≥ 0.7 or
≤ -0.7 are indicated with an asterisk. The grey highlighted variables are the variables used for the species distribution models.
Data preparation
Removing the missing data in the location data of the common Dandelion resulted in a dataset of 667 points, representing 230 diploids and 601 triploids. The Cuckoo flower data contained 391 locations after missing data removal.
Figure 1 shows the common Dandelion location data of the diploids and triploids plotted in Europe, the red dots represent the diploids, the blue dots represent the triploids. The diploids points were mostly located in central Europe, while the triploids were also present in northern Europe. Overall the Dandelions were overly represented in Belgium, the Netherlands and Switzerland and Austria. After removing duplicates, the diploids group of the common Dandelion contained 227 locations and the triploids 590. The subsamples were for common Dandelion: diploids 37 and triploids 60.
Figure 1. Present data of the common Dandelion. Overview of the present data of the common Dandelion in Europe. Diploids are represented by the red dots, the triploids by the blue dots.
The Cuckoo flower cytotypes were classified in ten different ploidy levels as shown in table 3. The three groups made from the ploidy levels were diploids (n=243), mediumploids containing triploids to hexaploids (n=139) and
higherploids containing the septaploids and higher (n=36). The distribution of each group is shown in figure 2. The diploids (red) and the mediumploids (blue) showed a similar distribution but mediumploids were also found in Belgium and the Netherlands. Higherploids (purple) were mostly distributed in northern Europe with a few
exceptions in Spain and Austria. Most samples of the Cuckoo flower derive from Switzerland and Austria and the area where Slovakia, Ukraine, Hungary and Romania meet. After removing duplicates, the Cuckoo flower diploid group contained 239, mediumploids 132 and the higherploids 36 locations after removing duplicates. The Cuckoo flower subsamples were: diploids 16, mediumploids 22 and higherploids 9.
Table 3. Overview of the grouping of the Cuckoo flower cytotypes in ploidy levels and the number of presents point within
each group. Ploidy 2n= x # individuals Diploid 16, 17, 18, 19 243 Triploid 20, 21, 24, 28 9 Tetraploid 30, 31, 32, 34, 36 75 Pentaploid 38, 39, 40, 44 60 Hexaploid 45, 46, 48, 52 9 Septaploid 53, 54, 56, 58, 59 30 Octoploid 60, 62, 64, 66, 67, 68 8 Nonaploid 69, 70, 72, 74, 76 7 Decaploid 78, 80, 84 3 Higher 118 1
Figure 2. Present data of the Cuckoo flower. Overview of the present data of the Cuckoo flower in Europe. Diploids are represented by the red dots, mediumploids by the blue dots and higherploids by the purple dots.
Species distribution models
Most models had AUC-values between 0.7 and 0.9, AUC higher than 0.9 were mostly found by GLMs. Bioclim and SVM most commonly had AUC-values below 0.7. These lower AUC were most often between 0.6 and 0.7 but in very few cases AUC around 0.5 were found.
For both species, all species distribution models using the four different future climate scenarios distribution did not seem to change drastically, covering most of the areas also covered by their current distributions. Most often the habitat suitability values got higher in already suitable areas, but in two cases habitats became unsuitable. Common Dandelion models
The predicted current distribution of Dandelion diploids was mostly centred around Central Europe. The distribution also included a large part of the Netherlands and a part of eastern Europe. East Germany was not included in its distribution. The distribution seemed to stop at Denmark, northern Spain and the farthest east it reached was Carpathian mountains (see figure 3a). The predicted distribution for the triploids of the common Dandelion match the distribution of the diploids, although the distribution of the triploids extended to further into northern Europe, including Denmark and the United Kingdom. It also included a large part of Poland in contrast to the diploids (see figure 3b).
Figure 3. Species distribution models of the common Dandelion. Combined species distribution models of the common Dandelion cytotypes using the AUC as a weighted mean. The colour indicates habitat suitability ranging from 0 to 0.20. a) Combined distribution model of the diploids using the AUC as a weighted mean. b) Combined distribution model of the triploids using the AUC as a weighted mean.
The diploid Dandelion was for all SSP’s distributed in Central Europe, but expanded in some models. Differences were especially visible in the model for SSP 585, which is the most extreme scenario, compared to the current distribution model. For SSP 585 diploids seemed to have moved east, although habitat suitability was only around 0.10. Here we also saw a decrease in habitat suitability in Italy (figure 4b). Overall habitat suitability for the diploids increased in the model for SSP 585 and SSP 370, mostly in northern and eastern parts of central Europe. For SSP 126, shown in figure 4a, habitat suitability increased in Belgium and northern France.
Present data of the Cuckoo flower in Europe
Figure 4. Future species distribution models of the common dandelion diploids. Combined species distribution models of the common Dandelion diploids using the AUC as a weighted mean. The colour indicates habitat suitability ranging from 0 to 0.20. a) Combined distribution model of the diploids showing the distribution for SSP 126. b) Combined distribution model of the diploids showing the distribution for SSP 585.
In the models triploids of the common Dandelion showed the most change (see figure 5). For SSP 126, triploids seemed to be distributed throughout Poland and triploids reached the border between Ukraine and Belarus (see figure 5a). For SSP 245 and 370 distribution expanded further along this border compared to the model for SSP 126. Distribution also expanded toward Lithuania (see figure 5b and 5c). For SSP 370 habitat suitability in Italy decreased (see figure 5c). In figure 5d the model for SSP 585 is visualized. Habitat suitability in Lithuania had increased compared to the other pathways and distribution had spread even more east. The triploids distribution for SSP 585 have expended Belarus and half of the Ukraine. The triploids habitat suitability in Italy was for this pathway decreased to between 0 and 0.05, except for the most northern part of Italy.
Figure 5.Future species distribution models of the common Dandelion triploids. Combined species distribution models of the common Dandelion triploids using the AUC as a weighted mean. The colour indicates habitat suitability ranging from 0 to 0.20. a) Combined distribution model of the triploids showing the distribution for SSP 126. b) Combined distribution model of the triploids showing the distribution for SSP 245. c) Combined distribution model of the triploids showing the distribution for SSP 370. d) Combined distribution model of the triploids showing the distribution for SSP 585.
Cuckoo flower models
The model for the diploid Cuckoo flowers showed that they were mostly distributed around Austria and the Carpathian mountains, and were likely to be distributed along the eastern coast of the Mediterranean sea and along the coast in southern France (see figure 6a). The mediumploids seemed to follow the same distribution as the diploids although distribution was less concentrated by the Carpathian mountains. Thereby mediumploids seemed to
a)
a) b)
b)
be more prominent along the French coast compared to the diploids. Unlike the diploids, mediumploids were likely to be distributed in Italy and northern Spain (see figure 6b). The higherploids, on the other hand, did not appear to have an overlap in distribution with the diploids and mediumploids. Higherploids were distributed in southern UK and the coast of northern France (see figure 6c).
Figure 6.Species distribution models of the Cuckoo flower. Combined species distribution models of the Cuckoo flower cytotypes using the AUC as a weighted mean. The colour indicates habitat suitability ranging from 0 to 0.20. a) Combined distribution model of the diploids using the AUC as a weighted mean. b) Combined distribution model of the mediumploids using the AUC as a weighted mean. c) Combined distribution model of the higherploids using the AUC as a weighted mean.
In the future models of the diploid Cuckoo flowers distribution stayed in the same areas but changed in habitat suitability. For SSP 126 distribution ranged in both the area around the Alps and the Carpathian mountains shrink (see figure 7a). Figure 7b shows that the distribution for SSP 585 was the same as for SSP 126 in the Alps, while the habitat suitability in the Carpathian mountains decreased. The distribution was projected to move to countries bordering the Mediterranean sea: Croatia, Bosnia and Herzegovina and Montenegro.
Figure 7.Future species distribution models of the Cuckoo flower diploids. Combined species distribution models of the Cuckoo flower diploids using the AUC as a weighted mean. The colour indicates habitat suitability ranging from 0 to 0.20. a) Combined distribution model of the diploids showing the distribution for SSP 126. b) Combined distribution model of the diploids showing the distribution for SSP 585. The distribution of the Cuckoo flower mediumploids for SSP 126 was similar to diploid distribution for SSP 126. But both the Alps and the Carpathian mountains distribution seemed to shrink and habitat suitability decreased (see figure 8a). The habitat suitability in Spain and southern France also decreased. For both SSP 245 and 370 did not change compared to the current distribution of the mediumploids. For SSP 585 habitat suitability in the Alps and Carpathian mountains decreased slightly (see figure 8b). However, habitat suitability in Spain and southern France increased.
a)
a) b)
c)
Figure 8.Future species distribution models of the Cuckoo flower mediumploids. Combined species distribution models of the Cuckoo flower mediumploids using the AUC as a weighted mean. The colour indicates habitat suitability ranging from 0 to 0.20. a) Combined distribution model of the mediumploids showing the distribution for SSP 126. b) Combined distribution model of the mediumploids showing the distribution for SSP 585.
The higherploids showed the most differences in distribution as a result of climate change (see figure 9). The model for SSP 126 showed an increase in habitat suitability in the UK and along the coast of the Netherlands and northern France. Furthermore, the distribution area had increased reaching Germany, Denmark and the coast of Poland. In northern Spain and southern France new distribution areas have formed (see figure 9a). The model for SSP 245 showed more resemblances to the current distribution model with a slight decrease in habitat suitability in the UK (see figure 9b). For SSP 370, there was an increase in habitat suitability in the UK and along the coast of the
Netherlands and northern France (figure 9c). The following places also showed a slight increase in habitat suitability, although less than the UK’s increase: the coast of the Netherlands and northern France, Denmark, the most southern part of Germany, southern France and along the coast of Poland, Lithuania, Latvia and Estonia. Surprisingly, the model for SSP 585 showed the same distribution as the current model (see figure 9d).
Figure 9.Future species distribution models of the Cuckoo flower higherploids. Combined species distribution models of the Cuckoo flower higherploids using the AUC as a weighted mean. The colour indicates habitat suitability ranging from 0 to 0.20. a) Combined distribution model of the higherploids showing the distribution for SSP 126. b) Combined distribution model of the higherploids showing the distribution for SSP 245. c) Combined distribution model of the higherploids showing the distribution for SSP 370. d) Combined distribution model of the higherploids showing the distribution for SSP 585.
a)
a)
b)
b)
Discussion
The models were in line with literature, both the diploids and triploids of the common Dandelion resided in Central Europe and triploids were also present in northern Europe. The future models showed that, if no actions were taken against climate change, both cytotypes would move away from the most southern parts of Europe and into eastern Europe. The distribution models of both the diploids and mediumploids of the Cuckoo flower showed that they were distributed along the Alps and Carpathian mountains, but mediumploids were also present at the border of Spain and in southern France. Higherploid distribution models had very low habitat suitability values but showed that they occurred in the UK and along the coast of northern France. In the future cytotype distributions of the Cuckoo flower would not change drastically. Only if no action was taken, diploids would become more present in Croatia, Bosnia and Herzegovina and Montenegro. Mediumploids would become more present in northern Spain and southern France. The higherploids models showed the most differences in distribution, although distribution was mostly centred in the UK and the coast of France and the Netherlands. All in all, climate change did not seem to have the effects expected at the beginning of the research.
The common Dandelion is already known to be present in eastern Europe, although research on the cytotype distribution there is lacking. The change in distribution to more eastern parts for both common Dandelion cytotypes in the models might be explained by bioclimatic variable 4 or temperature seasonality. In the future, temperature seasonality will decrease in eastern Europe, meaning that the temperatures in winter and summer will differ less from each other. This causes the climate in eastern Europe to closer represent that of central and northern Europe, where most sampling has taken place. The eastern Europe climate in the current situation is deemed not suitable by the model since almost no samples have been taken there. The shift to a climate closer resembling central and northern Europe causes the model to now include eastern Europe in the distribution.
It was expected that the diploid Dandelion would expand northward due to rising temperatures, the models, however, did not show this change in distribution. What they show was that both diploid and triploid Dandelions are likely to move away from the most southern part of their distributions. This could be explained by the fact that droughts will become more frequent and more intense in this part of Europe. Dandelions might not be resistant against such dry periods and therefore move away.
The Cuckoo flower date resulted in models that do not represent the distribution accurately, which is most likely caused by the sampling bias present in the literature. First of all, Both the diploids and mediumploids are distributed along the Alps and Carpathian mountains. This could be because almost all sampling was done in and around these mountain ranges making the sampling too biased to use for an accurate species distribution model. Points in Spain and France for both diploids and mediumploids are strongly underrepresented in the distribution models, likely due to the small number of samples in these areas compared to the Alps and Carpathians. Due to this sampling bias, the mountain climate, which differs quite a lot from the rest of Europe, is overly represented causing the exclusion of other climates such as those in Spain and France.
The low habitat suitability values and small distribution area of the higherploids might be due to the relatively small sample. This causes every point to have more influence on the model. It could be that only one or two location have, for example, an increase temperature it would change the temperature for suitable habitats more in a sample of 30 locations then for a sample of 130 locations. A slight difference in the variables of a location could cause a big difference in distribution, which was probably the case for the SSP 126 model. This model is based on climatic variables probably closest to the current situation compared to the other pathways. Therefor it would be expected that this would not differ much from the current situation. The model on the other hand shows the biggest difference in distribution both in the spatial sense and in habitat suitability. All in all, due to big sampling biases contained in the literature about the Cuckoo flower, the meta-analysis does not supply the right data to perform species distribution modelling accurately.
All models extracted from the meta-analysis show a certain degree of patchiness, this could be explained by two reasons. One, the models used were too sampling-site specific and therefore excluded more areas than needed. Two, the sampling sites were also quite patchy and therefore might create a more patchy model. The values for the habitat suitability returned by the models were also quite low (0 to 0.20), this is probably due to the combining of models, including some models with low predictive values. For a follow-up study, the use of one or two models might be preferable, or one could compare different models without combining them. For this data set the best model would be GLM since these models consistently had high AUC, the second best model was the random forest model. A flaw in the used method for species distribution modelling, especially for the common Dandelion, is the usage of presence points only. The common Dandelion cytotypes are present in mixed populations, meaning that frequencies of each cytotype within a population might also be of importance in the distribution. In the current modelling
methods all diploid and triploid presence points are weighted equally, while in practise they should be weighted by their respective frequencies.
Previous research about the differences between the diploid and triploid Dandelions mainly focussed on
thermophilic tendencies of the diploids as the main explanation of distribution differences (Roetman et al., 1988; Elzinga et al., 1987; Verduijn et al.,2003). In this research the thermophilic characteristic does not seem to play a large role in the distribution. The theory of easier colonisation by the triploids may need to be more experimentally explored to see if this indeed plays an important role in the current distribution.
For the Cuckoo flower the most common cytotype in Europe should be hypotetraploids (Lihova etal., 2003). However, in the literature study, the most observed cytotype is 2n=16 or diploid, with the hypotetraploids the secondly most frequent. This difference could be due to the sampling bias. If diploids are the most present in the sampled mountain areas, it would make sense that diploids are mostly found in this research since other areas are very underrepresented. Higherploids are found in the UK which is consistent with previous findings, although this only concludes a small part of their distribution as noted by Lihova et al. (2003).With more and wider spread sampling, distribution throughout the whole of northern Europe might be concluded. Diploids are present in central Europe, as found before (Lihova et al., 2003), but it would seem they are only in a small area of Central Europe. This might be because the sample location within the literature are bias and might disfigured the real distribution. Although this study did not find new insights in distribution patterns for cytotypes of different species, it does give insights in the available data. In future research, focus should be on requiring a more representative sampling area. For the common Dandelion this would consist of southern and eastern Europe and Scandinavia. For the Cuckoo flower most of Europe is still uncharted territory. For this project, studying the distribution of the cytotypes for the Netherlands was originally planned for both species, but these plans had to be changed because of the Covid-19 lockdown.
To study the distribution of the common Dandelion a better technique should be used that does include frequency. Unfortunately, techniques for species distribution models that take frequency in account are barely available. Distribution maps using frequencies will give more accurate results of the current distribution and therefore will also give better future models.
Since for the common Dandelion it might be possible that reproduction plays an important role in the distribution of the cytotypes, of course it would be interesting to investigate this aspect more. It would also be interesting to see if the higherploids use vegetative reproduction more than diploids because of their environment or if it is genetically determined. If it is genetically determined it maybe be linked to the different distribution patterns of the cytotypes. But before this it might be better to focus on studying the cytotype distributions in low sampled areas and with this look for differences in ecological preferences.
To conclude, climate change is not expected to drastically change the already known distribution of the cytotypes of the common Dandelion and the Cuckoo flower. Thermophilic tendency may not be the main driver in distribution differences of the diploid and triploid Dandelions as formerly thought. A meta-analysis is not a sufficient way for collecting location data for species distribution modelling of the Cuckoo flower due to the sampling bias present in the literature. In future research more attention should be paid to new sampling locations such as southern Europe, eastern Europe and Scandinavia. Also future research may focus on the role of reproductive strategies in the distributional differences of the different cytotypes.
references
Asker S.E. and Jerling, L. (1992). Apomixis in plants. Boca Raton, CRC Press.
Berg C. C. (1967) Cytotaxonomic studies in Cardamine pratensis L. in the Netherlands. Acta Bot. Neerl. 15: 683–689.
Castro, M., Loureiro, J., Serrano, M., Tavares, D., Husband, B.C., Siopa, C. and Castro, S. (2019). Mosaic distribution of cytotypes in a mixed-ploidy plant species, Jasione montana: nested environmental niches but low geographical overlap. Botanical Journal of the Linnean Society, 190, 51–66.
Cutler, F. and A., and Wiener, R. port by A. L. and M. (2018). randomForest: Breiman and Cutler’s Random Forests for Classification and Regression (4.6-14) [Computer software]. https://CRAN.R-project.org/package=randomForest
Elith, J. and Leathwick, J. R. (2009). Species Distribution Models: Ecological Explanation and Prediction Across Space and Time. Annual Review of Ecology, Evolution, and Systematics, 40(1), 677–697. https://doi.org/10.1146/annurev.ecolsys.110308.120159
Elzinga, D., Van Der Kamp, J., Den Nijs, J. C. M. and Sterk, A. A.(1987). Cytogeography and ecology of diploids and triploids of Taraxacum section Taraxacum in South Limburg, Netherlands. Proceedings Koninklijke Nederlandse Akademie van Wetenschappen, Serie C 90, 431– 442.
Explainer: How ‘Shared Socioeconomic Pathways’ explore future climate change. (2018, April 19). Carbon Brief.
https://www.carbonbrief.org/explainer-how-shared-socioeconomic-pathways-explore-future-climate-change Hijmans, R. and Elith, J. (2012) Species distribution modeling with R
Hijmans, R.J., Phillips, S., Leathwick, J. and Elith, J. (2017) Package ‘dismo’. Available online at:
https://cran.r-project.org/web/packages/dismo/dismo.pdf.
Karatzoglou, A., Smola, A., Hornik, K., Australia (NICTA), N. I., Maniscalco, M. A., and Teo, C. H. (2019). kernlab: Kernel-Based Machine Learning Lab (0.9-29) [Computer software]. https://CRAN.R-project.org/package=kernlab
Kolář, F., Čertner, M., Suda, J., Schönswetter, P. and Husband, B.C. (2017). Mixed-ploidy species: progress and opportunities in polyploid research. Trends Plant Sci. 22: 1041–1055.
König, M. A. E., Lehtilä, K., Wiklund, C. and Ehrlén, J. (2014). Among-Population Variation in Tolerance to Larval Herbivory by Anthocharis cardamines in the Polyploid Herb Cardamine pratensis. PLOS ONE, 9(6), e99333. https://doi.org/10.1371/journal.pone.0099333
Lihová, J., Tribsch, A. and Stuessy, T. F. (2004). Cardamine apennina: a new endemic diploid species of the C. pratensis group (Brassicaceae) from Italy. Plant Systematics and Evolution 245: 69–92.
Lihova, J., Tribsch, A. and Marhold, K. (2003). The Cardamine pratensis (Brassicaceae) group in the Iberian Peninsula: taxonomy, polyploidy and distribution. Taxon, 52, p. 783–801.
Mártonfiová, L. (2006). Possible pathways of the gene flow in Taraxacum sect. Ruderalia. Folia Geobotanica, 41(2), 183–201. https://doi.org/10.1007/BF02806478
Maes, J., Barbosa, A., Baranzelli, C., Zulian, G., Batista e Silva, F., Vandecasteele, I., Hiederer, R., Liquete, C., Paracchini, M. L., Mubareka, S., Jacobs-Crisioni, C., Castillo, C. P. and Lavalle, C. (2015). More green infrastructure is required to maintain ecosystem services under current trends in land-use change in Europe. Landscape Ecology, 30(3), 517–534. https://doi.org/10.1007/s10980-014-0083-2
Meirmans, P.G., Calame, F.G., Bretagnolle, F., Felber, F. and den Nijs, J.C.M. (1999). Anthropogenic Disturbance and Habitat Differentiation between Sexual Diploid and Apomictic Triploid Taraxacum Sect. Ruderalia. Folia Geobotanica, 34(4), 451–469. JSTOR.
van der Meijden, R. (2005). Heukels’ Flora van Nederland. 23th edition. Noordhoff Uitgevers, Groningen.
Roetman, E., den Nijs, J.C.M. and Sterk, A.A. (1988). Distribution and habitat range of diploid, sexual dandelions (Taraxacum section Vulgaria), a Central European flora element in The Netherlands. Acta Bot. Neerl. 37(1), p. 81-94.
Verduijn, M. H., Van Dijk, P. J. and van Damme, J. M. (2004a). The role of tetraploids in the sexual-asexual cycle in dandelions (Taraxacum). Heredity 93, 390–398 .
Verduijn, M.H., van Dijk, P.J., van Damme, J.M.M. (2004b). Distribution, phenology and demography of sympatric sexual and asexual dandelions (Taraxacum officinale s.l.): geographic parthenogenesis on a small scale. Biol. J. Linn. Soc. 82: 205–18
Salisbury, E. (1965). The Reproduction of Cardamine pratensis L. and Cardamine Palustris Peterman Particularly in Relation to their Specialized Foliar Vivipary, and Its Deflexion of the Constraints of Natural Selection. Proceedings of the Royal Society of London. Series B, Biological Sciences, 163(992), 321–342. JSTOR.
Stats package | R Documentation. (n.d.). Retrieved 12 July 2020, from https://www.rdocumentation.org/packages/stats/versions/3.6.2 Stocker, T.F., D. Qin, G.-K. Plattner, L.V. Alexander, S.K. Allen, N.L. Bindoff, F.-M. Bréon, J.A. Church, U. Cubasch, S. Emori, P. Forster, P. Friedlingstein, N. Gillett, J.M. Gregory, D.L. Hartmann, E. Jansen, B. Kirtman, R. Knutti, K. Krishna Kumar, P. Lemke, J. Marotzke, V. Masson-Delmotte, G.A. Meehl, I.I. Mokhov, S. Piao, V. Ramaswamy, D. Randall, M. Rhein, M. Rojas, C. Sabine, D. Shindell, L.D. Talley, D.G. Vaughan and S.-P. Xie, 2013: Technical Summary. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T.F., D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
Thuiller, W., Lavorel, S., Araujo, M. B., Sykes, M. T.,and& Prentice, I. C. (2005). Climate change threats to plant diversity in Europe. Proceedings of the National Academy of Sciences, 102(23), 8245–8250. https://doi.org/10.1073/pnas.0409902102
Appendix 1
Overview of the Cuckoo flower cytotypes and the amount of presence locations found in the literature.
# chromosomes # presence 16 272 17 11 18 14 19 6 20 1 21 2 24 7 28 2 30 80 31 1 32 24 34 1 36 2 38 20 39 1 40 6 44 34 45 1 46 4 48 6 52 1 53 1 54 3 56 24 58 3 59 2 60 4 62 2 64 2 66 1 67 1 68 6 69 1 70 1 72 3 74 4 76 6 78 3 80 1 84 1 118 1
