The role of ecotypic variation in driving worldwide colonization by a cosmopolitan plant

Short-living plants meet their main energy demand by photosynthesis of rosette leaves. These leaves optimize their yield as required for successful growth in so many different locations around the world. Do they cope with varying conditions by one general-purpose genotype and phenotypic plasticity or by ecotypic differentiation with a specific genetic background? How did leaves of a plant like Shepherd’s Purse, a weed colonizing all continents and climatic regions during a time period of only 500 years, evolve? We demonstrate that a combination of physiological, anatomical and morphological differentiation might allow such tremendous colonization success.


Introduction
Since Turesson's pioneering work (1922aTuresson's pioneering work ( , b, 1930, the ecotypic differentiation of plant species has been a main interest in population biology: how and how quickly are plant species able to adapt to changing environmental conditions. Due to their property of self-fertilization, short-living colonizing plant species are suitable examples. A single seed is able to colonize a new habitat and to establish a new population. Which traits play a role in this highly successful colonizing history? For a shortliving plant, several characters are critical, in particular, the physiology of germination and the determination of flowering time. In both developmental steps, the plant quits a status that is resistant to harsh conditions and changes to a highly sensitive status. If the environment is unfavourable, the individual will not complete its life cycle or it will produce only a low amount of immature seeds. Between germination and flowering, when they grow vegetatively in a rosette status, these plants are able to survive extreme conditions, e.g. cold winter. So far, numerous investigations have aimed at the control of flowering time, but less information is available concerning ecotypic differentiation with regards to a combination of morphological, anatomical and physiological leaf characters, which we are focussing on in this study. The Shepherd's Purse Capsella bursa-pastoris (Brassicaceae) belongs to the most prevalent flowering plants on earth (Coquillat 1951;Zhou et al. 2001), but it is not found in the hot and humid tropics. Their extraordinary colonizing success may be caused by the predominantly selfing mating system, rapid propagation by seeds as an annual to winter annual, the production of an enormous amount of seeds per individual (Hurka and Neuffer 1991), the ability to survive in a soil seed bank for many years (Hurka and Haase 1982) and the power for long-distance dispersal via myxospermy (Neuffer and Linde 1999).
In prehistoric times C. bursa-pastoris was distributed over the whole of Eurasia including the regions surrounding the Mediterranean Sea either along river shores or by early agricultural activities of humans. Later on, from the beginning of the 16th century, Europeans colonized all other continents, used the same agricultural techniques and crop plants and introduced many weeds as neophytes that in some cases turned out to be pests for the native biodiversity (e.g. Mooney et al. 2005). This unintentional transport paved the way for C. bursapastoris to reach the New World, Australia, South Africa, New Zealand, the Falkland Islands and other localities (Neuffer and Hurka 1999;Neuffer et al. , 2011Kryvokhyzha et al. 2016). This fast expansion of one weedy plant species was only possible due to its extraordinary capability of ecotypic differentiation. The differentiation of C. bursa-pastoris has been recorded for Europe (e.g. Bartelheim 1989, reviewed in Neuffer 2011), and such preadapted ecotypes have been able to find their niche elsewhere on the globe (Neuffer and Hurka 1999;. In many studies, the ecotypic differentiation of C. bursa-pastoris has been demonstrated for various regions of the world predominantly regarding germination, flowering time, rosette diameter and number of inflorescence branches (reviewed in Neuffer et al. 2011). Adaptive traits are frequently related to the rosette leaves that are responsible for the production of resources required for subsequent yield and abundant seed set. In the case of winter annuals, a robust rosette is required to survive suboptimal weather conditions. Important leaf traits are rosette diameter, the number of leaves in a rosette, leaf area, thickness of the leaf as well as of the epidermal cells, stomatal density, and photosynthetic capacity, as well as photosynthetic light utilization. Stomatal density and other epidermal characteristics strongly influence water-use efficiency (WUE), which is particularly important in dry habitats accompanied by high irradiation (reviewed in Körner 2003). The photosynthetic capacity depends on number, total area and anatomy of leaves and on cellular physiological parameters; in particular, the ability to dynamically adjust photosynthetic capacity to changing environmental conditions results in higher fitness (Athanasiou et al. 2010). The climatic adaptability and ecotypic intraspecies differentiation in a combination of morphological, anatomical and physiological characters have been shown for Diplotaxis erucoides populations from Sicily by Schleser et al. (1989).
The degree of leaf-margin dissection-from entire leaves with smooth margins to serrated and increasingly deeply lobed leaves-is likely to be functionally important. Leaf-margin dissection shows a very robust negative correlation with mean annual temperature, both at the within-species and at the community level. Such a wide-spread relationship between a morphological trait and an environmental parameter across different phylogenetic scales provides a strong argument that the trait in question is adaptive (Nicotra et al. 2011). Recently, the analysis of leaf size and compound leaves of a large number of species in relation to geography and climate was analysed by Wright et al. (2017) which is an indicator for global climatic change. The genus Capsella shows a high level of variation in all of the above-mentioned leaf-related traits, including the leaf shape, which ranges from entire leaves to very deeply dissected ones (Fig. 1;Shull 1909;Sicard et al. 2014).
Here, we will shed light on a combination of morphological/anatomical and physiological variation that underlies rapid local adaptation in one of the world's most successful weeds.

Methods
For each of the four data sets different treatments and determinations were performed as described below. It was our aim to analyse all leaves at the same stage of development. Therefore, we started the leaf studies when the first flower bud appeared indicating the end of the vegetative phase. In the case of late flowering plants, we started no later than 3 months after sowing to avoid any senescence.

Analysis 1: isotope analysis
To test the ability of progenies from different environmental habitats to react on drought stress we grew sister individuals under different conditions and analysed δ 13 C values in combination with morphological and anatomical features (Fig. 2, green colour; Table 1). Plants were grown in a growth chamber with a daynight rhythm of 15:9 h with 15-25 °C. Progeny of individuals collected in the wild (Fig. 2) were divided into two groups, one group was kept under water stress conditions with a maximum of 10 mL water daily per 1-L pot, the other group with at least 30 mL daily, for non-stress conditions. Each individual was planted in 1-L substrate with sand and slightly fertilized turf in a proportion of 1:2. Each population and each condition were represented by up to five sister individuals. For each individual (three individuals for each population and treatment), cell size, stomata density and the percentage of the volume of mesophyll cells compared with intercellular space were measured 30 times. Furthermore, the δ 13 C values were determined as follows: dried leaf material was combusted in an excess of oxygen at ~1000 °C, and the resulting CO 2 used for isotope analysis using the MAT 250 mass spectrometer (Schleser and Poling 1980). Carbon isotope values of leaves are based on total organic matter rather than on a selected chemical compound such as cellulose. Several test measurements have shown that relative variations of the isotope content led to similar results for total organic matter and cellulose. Only the absolute values differ by 2 to 2.5 °/ 00 . Results are reported in terms of δ 13 C relative to PDB (Belemnite from the Pee Dee formation in South Carolina; Craig 1957). The δ 13 C value as the ratio between 13 CO 2 and 12 CO 2 was measured twice for each individual.

Analysis 2: chlorophyll fluorescence and CO 2 gas exchange analysis
To test whether progenies from various environmental habitats differ in their ability for their photosynthetic activity, we analysed chlorophyll fluorescence and CO 2 gas exchange in combination with morphological and anatomical features (Fig. 2, blue colour; Table 1). Of each accession (see Fig. 2) three individuals were grown in a growth chamber with 12-h photoperiod and 15 °C day and 5 °C night temperature. For each individual, the thickness of the leaf and the epidermis cells, stomata density and leaf area were measured. Using a FluorCam 800MF (Photon Instruments, Brno, Czech Republic), we determined different chlorophyll fluorescence emission parameters of the whole rosette: F m : maximum fluorescence emission in light; F 0 : ground fluorescence in light; F: fluorescence emission in light after light pulses of 6000 µmol m −2 s −1 (for 800 ms every 30 s). From these measured parameters, the photosynthetic light utilization was estimated (Schreiber et al. 1986;Genty et al. 1989;Scheibe et al. 2005;Hanke et al. 2009;Scheibe and Dietz 2012;Silva et al. 2012;Voss et al. 2013).
Furthermore, the CO 2 gas exchange was measured using the Lic400XT Portable Photosynthesis System (Li-Cor Biosciences, Lincoln, NE, USA). A/C i curves enabled us to calculate the efficiency of RubisCO to fix CO 2 under limiting conditions. To determine the photosynthetic capacity of the secondary reaction in relation to the specific light intensity, light saturation curves were recorded. From the obtained parameter values the quantum yield for CO 2 uptake and the light compensation point (LKP) could be calculated.

Analysis 3: CO 2 gas exchange analysis under different light stress conditions
In order to test the ability of progenies from different environmental habitats to light stress, we cultivated sister individuals under different conditions and analysed CO 2 gas exchange in combination with morphological and anatomical features (Fig. 2, purple colour; Table 1).
We used material from two very different vegetation zones, namely the boreal (1461, 1530) and the meridional (2069, 2072) climatic region to carve out the ecotypic adaptation of the leaves to different environmental conditions (Fig. 2). For each population, up to 49 individuals of the progeny of two individual plants collected in the wild were used. The material was sown in a growth chamber with 12-h photoperiod and 15 °C day and 5 °C night temperature. The material was then divided into four experimental groups: 7.5-h photoperiod, 20 °C, high light setting (800 µmol m −2 s −1 , Fig. 3 (100 µmol m −2 s −1 , Fig. 3, right); 12-h photoperiod, 20 °C, medium light setting (150 µmol m −2 s −1 ); 12-h photoperiod, 15 °C day and 5 °C night ('cold'), medium light setting (100 µmol m −2 s −1 ).
The anatomical and physiological analysis was performed as in Analysis 2.

Analysis 4: thickness of the leaf in a New World transect
As the thickness of the leaf is not only a general character in adaptation to sunny or shady orientation of leaves but seems also to be a character for ecotypic differentiation within Shepherd's Purse, we performed a large New World transect including populations from South Africa (Fig. 2, orange colour; Table 1). The individuals were grown in a common garden field experiment and planted randomly in the Botanical Garden in Osnabrueck (Germany, May to July 2015). For anatomical analysis, material was taken directly from the field and stored in 70 % alcohol. After 1 day in tap water, the leaves became sufficiently soft for anatomical cuttings. The thickness of five rosette leaves as well as their upper and lower epidermis was determined for the terminal lobe of the leaf and for one lateral leaflet in one to two individuals of each population (Fig. 1). We decided to study different positions of the leaf as the information might differ; also the leaflets may differ between the leaf types (Fig. 1).

Statistical data evaluation
The data have been analysed statistically with the SPSS software package version 23. To test the normal distribution, we used the Kolmogorov-Smirnov test. In cases where data were significantly not normally distributed, we used the Spearman correlation for the correlation analyses; rho-value and significance are included in the figures. Only significantly correlated data are shown in the figures. The correlations are based on individual data. For testing significant difference (i) between treatments within a population or region, (ii) between populations within one treatment, (iii) between leaf types, we performed parameter-free Wilcoxon-test or the H-test of Kruskal and Wallis ( Table 2).
As in Analysis 4, when studying the leaf type of a progeny of 76 populations from various vegetation zones, we performed a post hoc Duncan test and an ANOVA.

Results
In this study, we correlate the results from the four described analyses with the different trait categories anatomy, physiology and morphology. A caveat in these analyses is, however, that we cannot ascertain the similarities of developmental maturation between the leaves analysed. Some of the differences observed, may, therefore, reflect intraspecific variation in life history.

Anatomical analyses
In Analysis 1, progeny from wild populations, when grown under water stress, developed denser mesophyll cells compared with the loose texture and large intercellular spaces in unstressed plants (Fig. 4). Furthermore, the palisade cells appeared narrower with a smaller Anatomical leaf parameters of various provenances grown under different conditions were correlated with geographical/elevational parameters at the places of origin (Fig. 5): the whole leaves and, in particular, the epidermis cell layer became significantly thinner with a higher degree of latitude (Fig. 5, all analyses, Fig. 6). In Analysis 4, we differentiated between the terminal leaflet (Fig. 5D1) and the lateral leaflet (Fig. 5D2) and observed at both positions that the thickness was the same. Interestingly, with a higher elevation at the place of origin leaves became thicker in Analysis 2 (Fig. 5C).
Even the epidermis layer itself varied with the degree of latitude and became thicker for populations originating from locations closer to the equator (Fig. 6). Stomata became less dense when populations originated from northern latitudes (Fig. 7). Cell sizes appear to decrease with the degree of latitude (Fig. 8, left) and increase with higher elevation (Fig. 8, right). However, only when grown under water stress conditions was the correlation highly significant.
Comparing populations originating from very divergent local conditions: with dry and hot conditions in the summer and high irradiation in Morocco, and temperate humid conditions and very long days in summer in Norway and in Karelia (Russia) (Fig. 9), the population from Morocco possessed thicker leaves and larger upper epidermis cells under all conditions, compared to the other populations. Whereas the Russian population showed a low variation of stomata density when grown under low or high light conditions and between cold versus warm temperature, respectively, the populations from Norway and Morocco increased stomata density when grown under high light and in cold conditions.  . Probability values: P < 0.05: significant differences (light grey); P < 0.001 highly significant differences (dark grey). Fig. 9

Physiological analyses
With water stress, the δ 13 C values were higher, meaning that more carbon isotopes had been fixed and assimilated during photosynthesis. As RubisCO discriminates the isotopes in the case of unhampered CO 2 uptake from the atmosphere into the intercellular spaces and across the cell membranes, higher values are a result of partially closed stomata and increase CO 2 isotope concentration within the leaf. With water stress, the proportion of cell volume to intercellular volume increased (consequently the intercellular space decreased, Fig. 10), and the stomata density also increased (see Fig. 7). Although the differences between water-stressed and non-stressed individuals are apparent in leaf anatomy (Fig. 4) and physiology (Fig. 10), leaf anatomy and δ 13 C values clearly differ between the provenances. The tenuis leaf type (provenance from the Alps) differed clearly from the other provenances by larger intercellular space compared to the cell volume. This trait coincided with lower δ 13 C values which might be the result of reduced ability to close stomata under water stress conditions. The difference between the δ 13 C values under water stress conditions might hint at an ecotypic differentiation with a high phenotypic plasticity for the provenances with the other leaf types (heteris, rhomboidea and simplex). The C. rubella individuals showed the highest stomata density without water stress. With water stress, these individuals intermingled in between the C. bursa-pastoris individuals, so that no differentiation between the two species was apparent from the characterized parameters. When excluding C. rubella from the analysis, only the correlation between the percentage cell/intercellular volume versus δ 13 C under water stress condition remained significant (Spearman rho = 0.635*). To substantiate differences between the two species, which are often to be found in mixed populations in their common distribution area, analysis of a larger number of populations is needed. The CO 2 -assimilation rate correlated highly significantly with the thickness of the leaf. Populations originating from higher latitudes develop thinner leaves under greenhouse conditions (Fig. 5B), enabling higher CO 2 -assimilation rates (Fig. 11). The non-photochemical quenching (NPQ) of the Russian population increased Figure 5. Spearman correlation of leaf thickness with latitude/elevation. The thickness of the leaf is significantly negatively correlated with latitude and in one case positively with the elevation; (A) = 25:15 °C, 9-h photoperiod, populations arranged according to the degree of latitude, three individuals have been tested for each population and treatment, both treatment groups differed significantly or highly significantly when tested by H-test of Kruskal and Wallis (see Table 2); (B and C) = 15:5 °C; 12-h photoperiod; (D) = common garden field experiment (n = number of individuals, in the figure are shown mean square values of the populations). Rho-value and significance (*α < 0.05, **α < 0.01) are included in the figures. A pairwise correlation test between the terminal and the lateral leaflet thickness (Kendall-, Friedmanand Wilcoxon-test) evidenced for significant differences between both. Only significantly correlated data are shown in the figures. The correlations are based on individual data. significantly under high light which might indicate higher stress from the increased temperature for these plants (Fig. 12). On the other hand, the NPQ of the Moroccan individuals is even lower under high light conditions, suggesting that these conditions are tolerated easily by these individuals (Fig. 12). The NPQ values are highly significantly negatively correlated with the stomata density at the lower surface (Spearman rho = 0.352**), namely, the stomata density increased at higher NPQ values. Efficient light use for CO 2 assimilation as can be recognized by photochemical quenching (qP) was highest in the Moroccan population under high light conditions, whereas the Russian population was characterized by low qP values (Fig. 12). Under all other environmental conditions, the populations displayed barely any differences. The qP values are significantly positively correlated with the stomata density at the lower leaf surface (Spearman rho = 0.284*) and with the area of a rosette leaf (Spearman rho = 0.281*).

Leaf types
The geographical distribution of the Mendelian leaf types according to Shull is apparent when regarding the measured leaf thickness from accessions along a transect through North and South America (Fig. 6): the simplex leaf type occurred more frequently close to the equator and seemed to be nearly absent at higher latitudes. This appears to be confirmed in the isotope analysis (Fig. 8), whereas, in order to be able to make a clear statement, the number of studied populations in Analysis 1 is too small. On  the other hand, in Analysis 2, this leaf type simplex did not correlate with latitude. In all analyses, the tenuis leaf type prefers temperate regions with adequate humidity during the vegetation period, even at higher altitudes (Analysis 1, Fig. 8). The reduced plasticity of tenuis compared with the other leaf types is also evident for the physiological traits, as the δ 13 C values were considerably lower compared with the values of the other leaf types, especially under water stress conditions (Analysis 1, Fig. 10). However, in Analysis 1, the number of analysed individuals and populations was restricted, and therefore, a higher sample number is necessary to verify our interpretation.
To summarize and generalize the results of our observations and experimental studies, the following statements are put forward: -The thickness of the leaf, of the epidermis and the epidermal cell size are negatively correlated with the degree of latitude.
-The stomata density varies significantly between different light conditions and provenances. -Physiological studies (δ 13 C values) showed that the leaf types/ecotypes heteris, rhomboidea and simplex appear to be able to close the stomata more efficiently under water stress conditions than the tenuis leaf type which might be due to lower plasticity. -Ecotypes with thinner leaves exhibit a lower maximal rate of CO 2 assimilation (A max ) at saturating light. -Physiological parameters resulting in high photosynthetic capacity under stressful, strong light conditions are typically found when the plants originate from hot, dry and sunny regions.

Discussion
The high degree of polymorphism of the leaves in the genus Capsella has been well known for more than Figure 8. Correlation of cell size with latitude (linear, Spearman) and altitude (quadratic). The correlation is highly significant under water stress conditions. Upper diagram: labels according to the leaf type; lower diagram: same correlation, but labels according to watering conditions (Analysis 1). Rho-value and significance (*α < 0.05, **α < 0.01) are included in the figures. Both treatment groups differed significantly when tested by Wilcoxon-test (see Table 2). Only significantly correlated data are shown in the figures. The correlations are based on individual data. 100 years. Almquist (1907Almquist ( , 1921 listed 200 elementary species, and in his opinion, this is a result of high variability of the genus in nature. In parallel, the geneticist Shull (1909Shull ( , 1911 performed extensive inheritance studies which formed the basis for the hypothesis of the existence of two Mendelian loci with two alleles, each responsible for the four basic leaf types within the genus. Later on, Shull argued in favour of an additional factor 'I' for leaves with completely entire margins ('simplissima', Shull 1929). Particular plants with small rosettes of linear leaves which have a spider-like appearance have been designated by Hus (1914) as xCapsella bursa-pastoris arachnoidea, and a leathery appearance corresponds to the dominant allele 'K' which was named 'coriacea' factor by Shull (1929). Clausen and Hiesey (1958) confirmed at least four pairs of genes responsible for the leaf shape in Capsella and suggested even a higher number of loci that are responsible for other modifications of the leaf. Our observations evidenced that, in the case of heterozygotes of tetraploid C. bursa-pastoris individuals, the dominance might be incomplete (e.g. AaaaBbbb) Figure 9. Anatomical data of provenances from different climatic/geographic regions under low (100 μmol m −2 s −1 ) versus high (800 μmol m −2 s −1 ) light conditions (upper part: 7.5-h photoperiod, 20 °C, Analysis 3), and cold versus warm temperatures (lower part: 12-h photoperiod, 150 µmol quanta/m 2 s, Analysis 3), three individuals for each provenance and treatment. All treatment groups differed significantly or highly significant when tested by Wilcoxon-test (see Table 2). Figure 10. Spearman correlation of δ 13 C values and anatomical traits. The δ 13 C values increased under water stress and were highly significantly correlated with anatomical traits under water stress conditions. The leaf type tenuis is characterized by lower δ 13 C values under both conditions compared with the other types (see 25:15 °C, 9-h photoperiod). Except for simplex leaf type (Capsella rubella), all individuals belong to Capsella bursa-pastoris. Rho-value and significance (*α < 0.05, **α < 0.01) are included in the figures. Only significantly correlated data are shown in the figures. Without C. rubella (phenotype simplex) only 'percentage cell/intercellular volume: δ 13 C' under water stress remained significant (Spearman rho = 0.635*). Both treatment groups differed significantly or highly significant when tested by Wilcoxon-test (see Table 2). The correlations are based on individual data. and then the leaf type would be intermediate and not distinguishable.
So far, we determined rosette leaves of many provenances worldwide (Fig. 13). In one case regarding provenances growing along an altitudinal cline, we detected an increase in the percentage of the B-allele with higher elevation (Neuffer and Bartelheim 1989). The B-allele is responsible for dividing the lobes to the midrib in the leaf types heteris and rhomboidea. In another analysis with provenances growing along a latitudinal cline, the variability in leaf types did not correspond to the northsouth gradient. In general, out of 15 050 scored leaves (Fig. 13), 19 % have been heteris, 51 % rhomboidea, 12 % tenuis, 11 % simplex and 7 % remained unscored.
The question is: why are the leaf types not randomly distributed in the case of selection neutrality and why is the percentage of the rhomboidea leaf type high? Are the leaf types adaptive by themselves, is the adaptation mirrored by anatomical or physiological characters, or is the leaf type linked to adaptive anatomical and/ or physiological leaf characters? Dissection of the leaf has been described as being inversely correlated with the mean annual temperature at the community and species level for trees (Royer et al. 2009). Therefore, dissection has been used for paleoclimatic reconstructions (Royer et al. 2005;Little et al. 2010), and it is conceivable that the dominance of the rhomboidea leaf type is a consequence of better adaptation of dissected leaves to various climates. In Capsella, the morphology of the leaf type depends for some reason on the environmental conditions. Capsella appears to exhibit earlier flowering times the longer the day in a long-day photoperiod (Hurka et al. 1976;Neuffer 1990). Under long day and warm temperature conditions some ecotypes flower so early that only a few rosette leaves are able to develop (Neuffer and Hurka 1986), resulting in these leaves which do not attain a pronounced leaf morphology but remain simple (Neuffer 1989). When grown under short day conditions and/or cold temperatures these provenances produce more rosette leaves (Neuffer and Hurka 1986), which enables them to reach the state which facilitates the development of the more pronounced leaf types (Neuffer 1989). In our study, the distribution of the leaf types was clearly divided into two subgroups according to the climax vegetation zone or the thermal vegetation zone according to Schroeder (1998) documented by the Duncan test in Analysis 4 (Table 3): regarding climax vegetation, only the leaf type tenuis belonged to a second subgroup, whereas regarding thermal vegetation zones, both tenuis and simplex comprised a second subgroup.
The molecular basis of the leaf shape in Brassicaceae is beginning to be unravelled. At first, the leaf shape seems to have evolved from small, simple leaves (Aethionema spec.) to compound leaves (Cardamine spec.). First results for Cardamine hirsuta have been obtained in the Tsiantis group: they hypothesize that 44 genes are potentially implicated in the leaf development, e.g. SHOOT MERISTEMLESS, BREVIPEDICELLUS or CUP-SHAPED COTYLEDON (Gan et al. 2016). For the leaflet formation in comparison with the simple leaves of Arabidopsis thaliana, the enrichment of the transcription factors of the PLETHORAS family is required, especially of PLT7 (Gan et al. 2016). Furthermore, Hay and Tsiantis (2016) detected a duplication of the gene LATE MERISTEM IDENTITY1 (LMI1) giving rise to REDUCED COMPLEXITY (RCO) in C. hirsuta. This duplication is lost again in A. thaliana and seems to be responsible for the reversal to simple leaves. In a detailed analysis of Sicard et al. (2014) with the two diploid species Capsella grandiflora and C. rubella, a second duplication which forms RCO-A and RCO-B has been detected. The difference between C. grandiflora with simple leaves (leaf type simplex) and C. rubella with dissected leaves (leaf type rhomboidea) was an allelic variation at the RCO-A locus. Furthermore, these authors detected four insertions of relatively recent origin in the RCO-A genomic organization which differed either in their absence or presence in various provenances. One future aim is to identify the molecular genetic background for the abovementioned, already known genes and alleles that model the morphology of Capsella rosette leaves. Finally, the adaptation of the rosette leaf is of highest importance, especially in the case of late flowering to biennial ecotypes overwintering with a rosette. In general, C. bursa-pastoris forms larger rosette leaves in later flowering plants under field conditions in common garden experiments (e.g. Neuffer and Hurka 1986;Neuffer 2011).
In this study, it is the first time that we report anatomical and physiological results of Shepherd's Purse leaf types. Körner (2003) reviewed leaf anatomical and physiological characters and discussed how leaves are adapted to high mountain ecosystems. He observed a significantly thicker mesophyll and larger epidermis cells for plants from higher altitudes which is in accordance with our findings. Regarding the fact that the climate in high latitudes of Scandinavia might be similar to high elevations in the Alps, the result for Capsella seems to be contradictory at first glance. However, in more northern latitudes, the days in the summer are longer and irradiation less strong. Therefore, the occurrence of thinner leaves with smaller epidermis cells in northern latitudes can be explained as a logical adaptation.
The physiological adaptation of leaves to various environments is often characterized by WUE, as can be deduced from δ 13 C values. In a comparison between different provenances of the grass Leymus chinensis from dry steppe regions of Asia, the differentiation under various conditions seemed to be more the result of plasticity rather than of ecotypic differentiation (Liu et al. 2016). The authors argue with the clonal propagation of this species which comes close to the general-purpose genotype in the sense of Baker (1974). In our case, the differentiation is apparently not only plastic but also ecotypic with a genetic background. We assume that ecotypes when growing under high light intensities at their places of origin are more adapted to high light, and are able to increase their quantum-yield efficiency considerably, whereas non-adapted genotypes are not able to do so or even suffer from photoinhibition as can be deduced from maximum quantum yield of PSII (F v /F m ) Figure 12. Chlorophyll fluorescence parameters of provenances from three different regions under low (100 µmol m −2 s −1 ) versus high (800 µmol m −2 s −1 ) light conditions (upper diagrams: 7.5-h photoperiod, 20 °C, Analysis 3), and low versus higher temperatures (lower diagrams: 12-h photoperiod, 150 µmol m −2 s −1 , Analysis 3). Parameters were measured under steady-state light conditions. F v /F m = maximum quantum yield of PSII; NPQ = non-photochemical quenching; qP = photochemical quenching (n = number of individuals tested). Most treatment groups differed significantly or highly significant when tested by Wilcoxon-test (see Table 2). values when analysed with the FluorCam. This ecotypic differentiation might be the result of the mixed mating system with an outcrossing of up to 12 % under good field conditions , whereas L. chinensis is a clonally propagating species (Liu et al. 2016).
Another aspect of the adaptation of the leaf are the qualitative and quantitative intraspecific variations of the main flavonoid pattern which was put forward by Eschmann-Grupe (1990) with populations of C. bursa-pastoris. The leaves appeared to reflect the adaptation of a population to the place of origin and varied with different environmental conditions. Five main and nine less prominent flavonoids were detected. Focusing on the main flavonoids the authors studied one population from high altitudes in the Alps, one from Norway and one from central Germany under various conditions in the growth chamber as well as in a reciprocal field experiment in 2000 m elevation in the Alps and in central Germany. The three populations varied qualitatively in their flavonoid composition only the population from the Alps contained all five main flavonoids. The Norwegian population contained no isoorientin, and the population from central Germany lacked diosmetin-7-O-β-D-glucoside. Under the various environmental conditions, the pattern did not differ qualitatively, but the quantity was increased significantly in field conditions, especially in the Alps. The composition and amount of secondary metabolites stored in the vacuoles of epidermal cells might be another physiological adaptation of Capsella ecotypes to high irradiation. These characteristics are possibly interesting for further studies of the ecotypic differentiation of Capsella rosette leaves besides the morphological, anatomical and photosynthetic parameters used in this study.

Conclusion
Here, we present a first insight into the ecotypic differentiation of Capsella rosette leaves in a combination of morphological, anatomical and physiological characters. The geographical distribution and frequencies of specific morphological leaf types seem to mirror the adaptation to particular environmental conditions at the places of origin. However, the actual adaptation might be overlaid  by anatomical and physiological adaptive traits which, with the numerous combinations of variations, point to a genetic background. To unveil the molecular background of the ecotypic differentiation of the Capsella rosette leaves, the knowledge of the molecular settings behind the leaf morphology is not yet sufficient. The clear geographic distribution pattern of the morphological leaf types might be partially adaptive by itself and therefore responsible for the frequency differences in various regions. However, the selection for specific morphological leaf types under different environmental conditions could be caused by genetic hitchhiking via anatomical or physiological adaptive characters with a genetic background. This linkage between the various traits is possible via closely linked loci on the same chromosome, and the effect is enhanced tremendously by the mating system which relies predominantly on selfing. It is, therefore, necessary to phenotypically and genetically elucidate all three aspects in combination: morphology, anatomy and physiology.

Sources of Funding
Financial support for expeditions to collect plant material was given in the framework of various projects funded by the DAAD, DFG and BMBF. Support for Analysis 4 was provided by DFG (NE 314/11-2).