An investigation into trait differentiation within and between two closely related Silene
species.
Daniel Connaghan
Degree project inbiology, Master ofscience (2years), 2017 Examensarbete ibiologi 45 hp tillmasterexamen, 2017
Table Of Contents
Abstract 3
Introduction 4
Materials and Methods 9
Collection sites and plant material 9
Trait Measurements 10
Statistical Analysis 13
Results 14
Climate Results 14
Between Species and Populations within Species Results 20
Climate Relationship Results 26
Discussion 27
Acknowledgements 32
References 33
Abstract
Ecological differentiation and adaptation are processes that can drive
divergence and speciation. Measuring ecologically revenant traits can help to identify possible targets of natural selection that may have mediated ecological differentiation.
This study looked for evidence of within and between species differentiation in seven ecologically relevant traits in two closely related species sampled across their range, and whether any of these traits were related to climate differences among site of origin.
We measured seven traits under common garden conditions in seedlings of Silene dioica (11 populations, n=528) and Silene latifolia (14 populations, n=672) grown in the botanical garden in Uppsala in a randomised block design. Three traits related to leaf morphology were measured, and four related to water usage of the plant were measured. These traits were analysed for differences between the species as well as for differences within each species between populations using a linear mixed model. The traits’ relationship to a climate variable, derived from the axes of a principal components analysis of climate data for each site, was investigated using a linear model.
A number of drought avoidance (e.g. succulence, max turgid weight) and morphological traits (e.g. leaf ratio) differed between the two species. Water use efficiency has been flagged before as possibly driving ecological differentiation between the two species and the results identify possible candidate traits for
quantifying this separation. Differentiation between populations within each species was also present in two traits within S. latifolia and in four traits within S. dioica significantly varying between the populations. This could reflect either local
adaptation or genetic drift acting on populations across the range. One trait related to the amount of water taken up by the leaf (wgain) was found to be significantly
associated with the climate variable, which was extracted from the principal components analysis, in S. latifolia.
The results identified a number of candidate traits which could reflect
ecological differences between the species especially with respect to water relations.
These traits may allow the species to respond differently during periods of water stress, which in turn could result in ecological separation of the species and determine their geographical ranges.
Introduction
Looking for evidence of trait divergence is an important step to understanding how speciation may occur, and at what stage in this process natural populations are.
Extrinsic (environmentally dependent) postzygotic isolation owing to differential adaptation and differences in ecogeographic distributions (Butlin et al., 2012) can reflect divergent ecological selection. The motivating idea behind the search for trait divergence between species is that gene flow between incipient species may be
reduced due to differential adaptions to the environment (Favre & Karrenberg, 2011).
This may then in turn lead to the evolution of intrinsic reproductive barriers between the two species, such as alteration in reproductive organs, or season in which they breed (Seehausen, 2014; Rahme, 2009 which act independently of the environment.
Traits that may contribute to ecological differentiation between species can vary largely within one or each species. It is important to also consider trait
divergence among populations of each species throughout their range. Trait variations within and between species measured in a common environment are generally
interpreted as genetically-based and can be explained either by natural selection driving phenotypic differentiation across climate conditions, by local selective pressures, or by genetic drift (Hall et al., 2007; Kawakami et al., 2011). Measuring traits important to plants’ survival across their range can provide an insight into which selective pressures are important for the plants and how these pressures vary in
importance across space.
In order to select relevant traits a number of important areas for plant ecology needed to be considered. Leaves are the main organs of the plant for carbon
assimilation and energy balance and thus a leaf’s functional significance in managing a whole host of stressors is obvious. Water use efficiency and specifically how plants deal with even moderate water stress is known as one of the most important forces governing range distribution (Taiz, 2014) and as such is worthy of investigation. In North America, vegetation analysis has shown that the distribution of plants is more correlated to traits related to the water relations of a plant, than with other commonly used metrics of climate (Stephenson, 1990), as changes in these water use traits more aptly reflects the effects of climate as sensed by plants (Stephenson, 1990). As well as water use traits, comparative investigations of traits related to leaf shape have also been indicated as useful in investigating adaptive significance of morphological
variation (Givnish, 1984). The shape of the leaf has been found to have relevance to the boundary layer resistance due to dead non-moving air surrounding the leaf (Gates, 2011) as well as follow on effects in terms of rate of transpiration (Chiariello, 1984).
Two-dimensional leaf shape has been shown to be strongly correlated to the structure of their veins, and broader leaves can potentially create stronger water gradients and higher photosynthetic function than found in narrower or more elongate leaves due to leaf tissue being positioned farther away from major veins (Leigh et al., 2014).
Utilizing lower photosynthetic function to avoid water lost to transpiration exceeding the water uptake is a well known drought strategy of plants (Roberts, 1986; Fitter, 1995), and thick leaves are characteristic of drought resistant xerophytic plants, as it gives a higher ratio of photosynthetic mesophyll to transpiring leaf area (Abrams et al., 1994). Studies investigating changes in leaf traits in relation to environmental gradients such as specific leaf area or leaf dimensions (Ackerly et al., 2002;
Ehleringer et al. 1981; Givnish, 1984) are relatively rare, but have found significant associations between species distribution and these leaf traits and have flagged the importance of further investigation of the changes in leaf traits across climate gradients (Ackerly et al., 2002).
In terms of candidate traits that were worthy of investigation, I looked for ones which had been shown in the past to be of biological relevance. With regards to traits related to water relations in plants, leaf succulence has been indicated as associated with a progressive inhibition of photosynthetic cellular machinery, as well as an inhibition of the key enzyme rubisco (Griffiths et al., 2007) and so plays a key role in how the plant photosynthesizes under variable light conditions. Succulent leaves usually possess large cells with thin walls and a large vacuole which makes them function well as water storage tissue (Taiz, 2014), and so this was identified as a possibly ecologically relevant trait keeping in mind the different life histories
regarding drought and shade tolerance of the two species discussed below. Maximum possible leaf turgidity as well as rate of water uptake has been shown to predict turgidity-dependent petiole flexibility which has knock on effects on the ability of a plant to resist wilting (Gonzalez-Rodriguez et al., 2016), and has previously been indicated as a candidate trait for drought tolerance in sugar beet (Ober et al., 2005).
Specific leaf area, as a measure of variation in leaf structure, was related to patterns of variation in net photosynthetic capacity in an analysis of field data of 107 species
species (Reich et al., 1998), and so was of interest to the plants which have been known to inhabit different light environments (see below). These traits provided a useful starting point from which to hone in on the study of potential ecological differences between the species.
This study focused on a number of aspects of trait divergence in traits in two closely related species of the plant family Caryophyllaceae. I looked at whether evidence for trait divergence could be detected in a collection of ecologically relevant traits including those related to water usage (eg. wgain, succulence, turgidw) and also leaf morphology (eg. SLA, leafratio) were investigated. The two study species, Silene dioica (L.) Clairv and Silene latifolia Poiret, have been model organisms for many past study questions in ecology and evolutionary biology and this wide breadth of literature made them an attractive model species pair for the study. There is existing research ranging from descriptions of life history and ecological observations for the genus to many more current studies investigating a wide group of evolutionary and genetic questions (Bernasconi et al., 2009). Both species are short-lived perennials (McNeill, 1978) which are fairly common throughout their distribution (Hathaway et al., 2009). The two species are sympatric throughout much of their range in central Europe (Hathaway et al., 2009), where chloroplast haplotypes are mostly shared between the two (Muir & Filatov, 2007). The range of S. latifolia spreads from the southwest of Europe throughout the continent and into Russia, and from the
Mediterranean to high latitude sites on the coast of Norway (Taylor & Keller, 2007).
It has also spread to North America (McNeill, 1978), where it is considered an invasive plant (Taylor & Keller, 2007). Silene dioica is distributed all across Europe, but is mostly absent from the Mediterranean (Hathaway et al. 2009).
The question of how S. dioica and S. latifolia remain distinct species
throughout their zones of sympatry persists, as despite being fully interfertile, early generation hybrids are quite uncommon (Karrenberg & Favre, 2008; Minder et al., 2007). Phenological differences in timing of flowering (Baker, 1947a), and
pollination syndromes (Bopp & Gottsberger, 2004; Goulson & Jerrim, 1997) between the two species have been indicated as possibly important intrinsic reproductive barriers in maintaining the species boundary between the two. Silene latifolia is visited by nocturnal pollinators (Goulson & Jerrim, 1997) while S. dioica is visited mainly by diurnal insect species (Bopp & Gottsberger, 2004). Another aspect of floral distinctions between the two plants is a partial pre-mating barrier involving
scent differences reducing pollen transfers between the species (Waelti et al., 2008).
Exactly how important this barrier is compared to the differences in the flowers between the two species is still undecided, but it is likely that it does play some role (Goulson & Jerrim, 1997). A post-pollination but pre-seed fertilization difference in the two species to heterospecific pollen success has been identified as an additional asymmetric barrier to hybridization in S. latifolia but not in S. dioica (Rahme, 2009;
Montgomery et al., 2010) and functions in a partial role limiting introgression
between the two species. However more hybrids are produced in S. latifolia than in S.
dioica in equally sized populations despite this pollination competition barrier (Rahme, 2009).
Differential adaptation has been pointed to as one of the main factors maintaining the species boundary between S. dioica and S. latifolia (Goulson &
Jerrim, 1997; Favre et al., 2016). Silene latifolia is found in drier conditions on wasteland with intermittent disturbance, while S. dioica occurs in conditions with more shade and in rich meadows (Baker, 1947a,b; Goulson & Jerrim 1997; Goulson, 2009; Karrenberg & Favre 2008). Various morphological differences between the two plants are consistent with ecological differentiation. Silene latifolia has narrower leaves than S. dioica, that can be up to twice as thick (Baker, 1948). Root systems are also very relevant to how a plant manages water. Silene latifolia has a deeper and thicker taproot in comparison to S. dioica’s thinner and shallower roots (Baker, 1947;
Friedrich, 1979), and this is in line with other studies investigating root architecture and its relevance in response to drought conditions (Ramamoorthy et al., 2017).
These features relate to differences in the plants’ usage of water, which can indeed be highly ecologically relevant, and could lead them to fill very different niches within sympatric zones (Silvertown et al., 2014). Studies have flagged water use efficiency and drought tolerance as possibly relevant separations between the species before, saying that water availability is lower in certain populations of the two species in S. latifolia habitats (Favre & Karrenberg, 2011), or that when grown in a greenhouse under drought conditions S. dioica starts wilting while S. latifolia is still at full turgidity (Baker, 1948). As mentioned before, S. latifolia is found in barer, drier conditions and S. dioica is commonly found in more humid forests and in dense overgrowth and, as such, water usage could potentially be an important trait separating their ecological niches and contributing to their range distributions.
The purpose of the present study was to investigate the differences in ecologically relevant traits within and between S. latifolia and S. dioica using a common garden experiment. Specifically, I asked the following questions: (i) Which traits differ between the two species? (ii) Which traits differ within each species across its range? (iii) Are any of these differences related to climate? After
answering these questions, we can discuss what it is possible to say about the nature of the species boundary between the two study species based on the analysis of the traits measured in this study.
Materials and Methods
Collection sites and plant material
We used maternal seed families of eleven populations of S. dioica and fourteen populations of S. latifolia that were sampled from field sites across the two species’ ranges, in Europe as well as two sites in Russia and one in Lebanon (Figure 1). One to six seed families from each of the populations were grown with ten replicate plants per family where possible, yielding a total number of 1200 plants.
(Table S1 for complete planting list).
Figure 1 Map of the species' (Silene dioica and Silene latifolia) distributions with field sites marked as dots, from unpublished work by Sophie Karrenberg.
Seeds were germinated in petri dishes on moist paper, and then grown for seven days under long day conditions (16h / 8h (day/night)) in a growth chamber. The germination rounds were staggered by two weeks in order to facilitate several rounds of trait measurements on the plants to take place at the same stage of development later in the experiment.
Where possible, eight healthy seedlings from each family in each germination round were potted in small pots in a mixture of potting soil (Weibulls) and clay beads (Leka beads, Weber). Plants were grown in a greenhouse at Uppsala University under 12 hour days. Seedlings were arranged in a randomized block design with eight blocks, each block receiving one individual of each family. Each block was made up of five trays and each tray contained thirty-six positions for the seedlings. The thirty remaining positions in each block, which were not allocated a seedling, were left empty.
For the first month, the seedlings were grown under green shading cloth. The plants were irrigated when necessary, usually once every three to four days, by filling the tables that the trays were situated on with water, as well as additional misting from the hose in the greenhouse.
Trait Measurements
Traits were measured on young, healthy plants after they had been situated uncovered in a strong light environment for two and a half weeks (Table 1). The sampling took place on a sampling schedule staggered by one week, with one half of the blocks being sampled at time point one, then the other half of the blocks’ sampling occurring a week later. The overall sampling period was approximately four months long, from December to March 2015. Half of the blocks’ plants were designated for more destructive sampling including water retention capacity and measuring specific leaf area. The variability between plants in response to removal of leaves was not a trait that we wanted to investigate, and was removed by separating the blocks into invasive and non-invasive sampling blocks. This approach was taken to ensure the destructive sampling would not have an effect on the non-invasive trait
measurements, as it has been shown (Marquis, 1996) that plants delegate resources after damage to under compensate, fully compensate, and over compensate for the damage received. Dry, pressed leaves were scanned and leaf area (cm2) was determined using ImageJ (http://rsb.info.nih.gov/ij/). Specific leaf area (SLA) was
calculated as leaf area/dry mass (cm2 g−1). Leaf succulence was calculated as (fresh mass - dry mass)/leaf area (g H2O cm − 2) (Reiman & Breckle, 1995). Other derived measurements (Table 2) were created using the simple measurements (Table 1), ending with a total of seven simple measurements and five derived measurements.
Any measurements of drought were all destructive and taken on the same leaf during a single round of sampling. I did not use the entire array of simple traits measured in the statistical analyses as many of them were strongly correlated with the derived measurements which were calculated using the simple trait measurements as well.
Other traits which were measured on the plants as well but were not further analyzed or presented, included seed weight per plant family, the number of leaves per plant for each round of sampling, the water loss of a leaf after 24 hours of drying, and the change in area of the largest leaf. A small pilot study on freezing damage to the leaves was also undertaken but not included as I didn’t think measuring freezing damage on leaf discs had much relevance to freezing tolerance in a complete alive leaf.
Every attempt was made to make the greenhouse a constant environment, and a randomized block design was used to limit the spatial variation in the greenhouse such as increased humidity, but it is possible that some contamination could have crept in. An example of this is how the sprinklers in the greenhouse were not uniformly distributed across the plants, and while turned off they were still dripping water and possibly affecting humidity in different parts of the greenhouse.
Climate Methods
Location and climate data of the field sites was extracted from WorldClim (resolution 1 km2, www.worldclim.org, Hijmans et al., 2005). The climate variable was made up of the PC1 axis of a PCA performed using the monthly temperature and precipitation averages for months which we decided were representative of the growing year.
Abbreviations used in the analysis are in Table 3.
Table 1 Description of the leaf measurements taken on Silene dioica and Silene latifolia seedlings.
Table 2 Summary table of the derived measurements calculated for Silene dioica and Silene latifolia.
Name Description Round of
Measurement length2 The length of the largest leaf on the plant, during the first measuring
round. Measured in mm using electronic calipers.
2 width2 The width of the largest leaf on the plant. Measured in mm, using
electronic calipers. 2
leafarea The area of the exact leaf (mm2), measured by outlining the leaf shape in ImageJ software and getting the number of pixels included by the leaf, then converting that to mm2 by using a scale.
1 & 2
freshw The fresh weight of the leaf, measured less than one hour after picking from the plant. Measured in mg on scales of accuracy to 0.01mg. After measurement the leaf was stored in a plastic bag covered with water to enable the measurement of turgidw the following day.
1 & 2
turgidw The weight of the leaf at full turgidity, after soaking in water overnight and weighed in mg on scales of accuracy to 0.01mg.
1 &2 daylater The weight of the leaf (mg) 24h after removing from the turgid state,
and after placing it in a well ventilated growth chamber to let it dry.
1 & 2 dryw The weight of the leaf (mg) on scales of accuracy to 0.01mg, after
drying in an oven on low heat for a minimum of eight hours.
1 & 2
Name Explanation
SLA Specific leaf area (mm2mg-1). From the measures above: leafarea divided by the dryw of the same leaf.
leafratio A ratio between width and length to show leaf shape. width2/length2.
Unitless.
weightd The change in the weight of the leaf (mg) from its fully turgid state, to the point a day later.
succulence The turgid weight (mg) of the leaf minus the dry weight (mg) of the leaf, divided by the leaf area (mm2) of the leaf (mg1mm-1).
wgain The weight the leaf (mg) gained from its fresh weight after reaching full turgidity. Turgid weight minus the fresh weight.
Statistical analysis
To determine the differences in the traits between the plants I used a variety of statistical approaches. Analyses were done using the most recent edition of R Studio version 0.99.491 which used R version 3.2.1 (RStudio Team 2015; R Core Team 2015).
I first approached the data set using multivariate analysis. A principal
components analysis (PCA) was performed using the climate data and the trait data in order to look at the underlying structure of the data, and notice any patterns in the data.
In order to look more closely at differences in the trait means between the two species, and between populations within each species, mixed models were used in the R packages ‘lmerTest’ (Kuznetsova et al., 2015) and ‘lme4’ (Bates et al., 2015). For the first comparison, which was between the two species, I treated population, family identity and block number as random effects in the model, and species was a fixed effect. The analyses were performed on the complete dataset including both species, as I wanted to look for differences between them.
For the second comparison, that of the different populations within each species, I used a mixed model where family identity and block number were random effects in the model with population as the fixed effect. I performed this analysis on each species’ data separately. If there were significant results returned for population, a Tukey post-hoc test was performed using the package lsmeans (Lenth, 2016) in order to see where the differences between populations occurred.
I used a simple linear model to test whether variation in population traits means within each species could be explained by climate, as PC1 scores of the climate analysis described above.
Table 3 Summary of the Abbreviations used in the WorldClim package.
Abbreviation Explanation
tmax1 The maximum temperature in the first month of the year.
tmin1 The minimum temperature in the first month of the year.
tmean1 The average monthly temperature in the first month of the year.
prec1 The amount of precipitation of the first month of the year.
Results
Climate Results
The sites were spread across a wide latitudinal gradient and their climate profiles varied widely. (Figure 1, Table 4). Averages of mean annual temperature and total annual precipitation across the sites ranged from -1.6 degrees (DOM1– S. latifolia) and 148.2 cm (LEKA1 – S. dioica), to 11.7 degrees (FRA1– S. latifolia) and 40.2 mm (ESP1– S. latifolia). Walter diagrams were created for each population site to best exhibit the yearly profile of temperature and rainfall.
The PC1 axis explained 65.51% of the variance in the PCA of the climate axis, and PC2 axis explained 17.83% of the variance.
Table 4 Location and climate data of the field sites extracted from WorldClim (resolution 1 km2, www.worldclim.org; Hijmans, 2005).
Pop Name
Country Nearest Town
Latitude Longitude Avg Annual Precipitation
(mm)
Longest Day Length
(h)
Avg Yearly
Temp (°c)
Species
AUS1 Austria Weiden am
See 47.91° 16.88° 51.197 15:23 12 S. latifolia
CH2 Czech
Republic
Bonaduz 46.81° 9.39° 97.392 15:35 10.3 S. latifolia
DOM1 Norway Otta 62.01° 9.5° 69.227 16:12 10.6 S. latifolia
ESP1 Spain Hoyo de
Manza 40.62° -3.894° 40.193 14:37 10.6 S. latifolia
FRA1 France Plouharnel 47.58° -3.10° 74.863 15:26 11.6 S. latifolia GB1 Great
Britain
Rattray 57.62° -1.87° 67.745 16:35 7 S. latifolia
GER3 Germany Helderbach 51.29° 11.22° 42.074 15:38 10.1 S. latifolia
GER4 Germany Allensbach 47.7° 9.10° 77.481 15:38 3.6 S. latifolia
LIB1 Lebanon Ariz 34.25° 36.03° 78.832 13:56 7 S. latifolia
POL2 Poland Gizycko 54.02° 21.68° 55.407 15:56 10.9 S. latifolia
POL3 Poland Torun 53.03° 18.60° 44.074 15:56 11.6 S. latifolia
RO3 Romania Micfalau 46.03° 25.82° 50.242 14:57 7 S. latifolia
RUS2 Russia Dvurechensk 56.6° 61.06° 41.685 16:32 8.7 S. latifolia
AB1 Sweden Abisko 68.35° 18.83° 64.058 22:41 -1.3 S. dioica
BEL1 Belgium Biemont 49.59° 5.57° 79.17 15:44 11 S. dioica
CH1 Czech
Republic
Klosters-
Serneus 46.84° 9.87° 82.23 15:35 13.4 S. dioica
CZ1 Czech
Republic Rakvice 48.86° 16.83° 47.906 15:29 10.8 S. dioica
GB2 Great
Britain Edinburgh 55.92° -3.17° 58.972 16:35 7.7 S. dioica
GB4 Great Britain
Sheffield 53.38° -1.53° 71.092 16:35 10 S. dioica
GER2 Germany Wechselburg 51.0° 12.77° 46.839 15:38 1 S. dioica
GER6 Germany Rietheim-
Weilheim 48.03° 8.75° 95.354 15:38 10.9 S. dioica
KON1 Norway Hjerkinn 62.30° 9.60° 68.458 16:52 10 S. dioica
LEKA1 Norway Leka Island 65.09° 11.75° 148.615 16:52 1.6 S. dioica
POL1 Poland Borowiec 54.41° 18.41° 48.146 15:56 10.6 S. dioica
RUS1 Russia Urmanchino 55.185° 58.61° 44.577 16:32 7.7 S. dioica
TRO1 Norway Tromso 69.67° 18.97° 77.32 22:21 10.1 S. dioica
Figure 2 Principal Components Analysis of the climate data measured for the sites.
Black points indicate Silene latifolia, and red points indicate Silene dioica.
Figure 3 The arrows (vectors) are pointing in the direction of the climate variables (explained in Table 3), as projected into the 2-d plane of the plot in Figure 2.
Figure 4 Principal Components Analysis of the means of all the trait data measured.
Black points indicate S. latifolia, and red points indicate S. dioica.
weightd.m leafratio.m
sla.m
succulence.m wgain.m
freshw.m
turgidw.m
Figure 5 The arrows (vectors) are pointing in the direction of the trait mean variables, as projected into the 2-d plane of the plot in Figure 4.
Trait variation between species and, among populations within species
The analyses returned significance for five traits for differences between species (Table 5), and two traits where there were no differences between species. Leaf ratio, weightd, succulence, turgidw, and freshw were all significant differences between the species. For definitions of these metrics refer to Table 1 and Table 2. The mean leaf ratio value for S. dioica was lower than that for S. latifolia. The weight difference for the leaves from S. latifolia was greater than that of leaves from S. dioica. Succulence was lower in S. latifolia than in S. dioica. Populations were significantly different for one trait in S. latifolia, and three traits in S. dioica. For S. dioica the traits that differed were succulence, fresh weight, and leaf ratio. Succulence was higher in S. dioica plants from populations in high latitudes. For S. latifolia the trait that differed was freshw. (Table 6, 7).
Table 5 ANOVA results table for the effect of species on a collection of ecologically relevant traits in Silene latifolia and Silene dioica. Significant results are marked in bold in the Trait column and indicates test significant at α =0.05.
Trait Variable Sum Sq Mean Sq
Num DF
Den DF
F Value
Pr (>F) weightd species 2.209 2.209 1 19.289 8.192 0.009 succulence species 236.68 236.68 1 19.478 9.889 0.005
wgain species 0.152 0.152 1 16.334 0.829 0.375
freshw species 9673.7 9673.7 1 21.043 6.074 0.022 turgidw species 21246 21246 1 21.111 8.478 0.008 leafratio species 3.738 3.738 1 20.289 84.013 1.18 ⋅10-8
sla species 0.181 0.181 1 18.686 3.655 0.071
Table 6 ANOVA results table for the effect of population on a number of ecologically relevant traits within Silene latifolia. Significant results are marked in bold in the Trait column and indicates test significant at α =0.05.
Table 7 ANOVA results table for the effect of population on a number of ecologically relevant traits within Silene dioica. Significant results are marked in bold in the Trait column and indicates test significant at α =0.05.
Trait Variable Sum Sq Mean Sq Num
DF Den
DF F Value Pr (>F)
weightd pop 3.139 0.314 10 150.73 0.894 0.541
succulence pop 0.119 0.012 10 106.95 1.912 0.05
wgain pop 2.954 0.295 10 155.02 1.354 0.207
freshw pop 42117.000 4211.700 10 154.03 2.359 0.012 turgidw pop 57850.000 5785.000 10 153.64 1.899 0.149 leafratio pop 39.361 3.578 10 59.765 113.260 2.2 ⋅10-16
sla pop 1.566 0.156 10 109.03 2.903 0.202
Trait Variable Sum Sq
Mean Sq
Num DF
Den DF
F Value
Pr (>F)
weightd pop 4.425 0.34 13 179.03 1.747 0.054
succulence pop 0.076 0.006 13 127.03 1.071 0.039
wgain pop 3.859 0.297 13 181.04 2.202 0.212
freshw pop 44979 3460 13 181.53 2.479 0.004
turgidw pop 59031 4541 13 181.28 2.200 0.111
leafratio pop 19.704 1.408 13 38.26 27.187 0.13
sla pop 0.936 0.072 13 127.05 1.668 0.075
(mg^1mm^-1) (mg^1mm^-1)
Figure 6 Populations means with standard errors for weightd (A,B) and succulence (C,D) in Silene latifolia (B, D) and Silene dioica (A, C). Populations are arranged in order of increasing latitude. Bars with different letters are significantly different, NS designates traits without significant differences between populations.
wgain (g)
0 10 20 30 40 50
CH1 GER6 BEL1 GER2 GB4 POL1 GB2 KON1 LEKA1 ABI1 TRO wgain (g)
0 10 20 30 40 50
LIB1 ESP1 RO3 CH2 FRA1 GER4 AUS1 CZ1 GER3 POL3 POL2 RUS1 RUS2 GB1
freshw (g)
0 50 100 150 200
CH1 GER6 BEL1 GER2 GB4 POL1 GB2 KON1 LEKA1 ABI1 TRO
a ab a a
b
ab ab
a ab
a ab
freshw (g)
0 50 100 150 200
LIB1 ESP1 RO3 CH2 FRA1 GER4 AUS1 CZ1 GER3 POL3 POL2 RUS1 RUS2 GB1
abac a abab
ab ab
ac ab
bc abab
ab b
E. F.
G. H.
NSA. NSB.
C. D.
(mg)(mg) (mg)(mg)
Figure 7 Populations means with standard errors for wgain (A,B) and freshw (C,D) in Silene latifolia (B, D) and Silene dioica (A, C). Populations are arranged in order of increasing latitude. Bars with different letters are significantly different, NS
turgidw (g)
0 50 100 150 200 250
CH1 GER6 BEL1 GER2 GB4 POL1 GB2 KON1 LEKA1 ABI1 TRO turgidw (g)
0 50 100 150 200 250
LIB1 ESP1 RO3 CH2 FRA1 GER4 AUS1 CZ1 GER3 POL3 POL2 RUS1 RUS2 GB1
leafratio
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
CH1 GER6 BEL1 GER2 GB4 POL1 GB2 KON1 LEKA1 ABI1 TRO
ab ab a
b ab abab c
ab ab abc
leafratio
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
LIB1 ESP1 RO3 CH2 FRA1 GER4 AUS1 CZ1 GER3 POL3 POL2 RUS1 RUS2 GB1
NS
NS NS
I. J.
K. L.
A. B.
C. D.
(mg) (mg)
Figure 8 Populations means with standard errors for turgidw (A,B) and leafratio (C,D) in Silene latifolia (B, D) and Silene dioica (A, C). Populations are arranged in order of increasing latitude. Bars with different letters are significantly different, NS
designates traits without significant differences between populations.
SLA
0 5 10 15
CH1 GER6 BEL1 GER2 GB4 POL1 GB2 KON1 LEKA1 ABI1 TRO SLA
0 5 10 15
LIB1 ESP1 RO3 CH2 FRA1 GER4 AUS1 CZ1 GER3 POL3 POL2 RUS1 RUS2 GB1
B.
B.B .
A. B.
NS NS
(mm^2mg^-1)
(mm^2mg^-1)
Climate Relationship
The results from the linear model for the two species (Table 8 and 9) only returned one trait that had a significant association to the climate variable, which was wgain in S. latifolia (Table 9).
Figure 9 Populations means with standard errors for SLA (A,B) in Silene latifolia (B) and Silene dioica (A). Populations are arranged in order of increasing latitude. Bars with different letters are significantly different, NS designates traits without
significant differences between populations.
Table 8 Results table for the association of climate (pc axis 1) on ecologically relevant traits in Silene dioica. Significant results are marked in bold in the Trait column and indicates test significant at α =0.05.
Table 9 Results table for the association of climate (pc axis 1) on ecologically relevant traits in Silene latifolia. Significant results are marked in bold in the Trait column and indicates test significant at α =0.05.
Trait Variable DF Sum Sq Mean Sq
F Value
Pr (>F) weightd pc1 1 0.023 0.023 0.502 0.496
Residuals 9 0.417 0.046 succulence pc1 1 0.001 0.001 0.35 0.568
Residuals 9 0.009 0.001
wgain pc1 1 0.015 0.015 0.464 0.512
Residuals 9 0.293 0.032
freshw pc1 1 0.001 0.003 0.35 0.568
Residuals 9 0.009 0.001 turgidw pc1 1 79.600 379.62 0.538 0.481
Residuals 9 6345.700 705.08 leafratio pc1 1 0.091 0.091 1.644 0.231
Residuals 9 0.499 0.055
sla pc1 1 0.006 0.006 0.395 0.545
Residuals 9 0.147 0.016
Trait Variable DF Sum Sq Mean Sq
F Value
Pr (>F)
weightd pc1 1 0.000 0.000 0.012 0.915
Residuals 12 0.430 0.036 succulence pc1 1 0.000 0.000 0.354 0.562
Residuals 12 0.010 0.001
wgain pc1 1 0.172 0.172 9.801 0.008
Residuals 12 0.210 0.018 freshw pc1 1 240.900 240.860 0.817 0.384
Residuals 12 3539.400 294.950 turgidw pc1 1 128.500 128.480 0.273 0.61
Residuals 12 5649.500 470.790 leafratio pc1 1 0.023 0.023 0.608 0.45
Residuals 12 0.456 0.038
sla pc1 1 0.036 0.036 1.997 0.183
Residuals 12 0.215 0.018
Discussion
This study focused on identifying potential differences in ecologically relevant traits between two species, and found seven traits that differed between species
(Figure 6, Table 5), and that one of the traits showed a significant relationship to the climate variable (Table 6, Figure 7). While the overall pattern in the data was that of large variation between the species and also within each population, the results are relevant in that they may point to ecological differentiation in traits between the species which could influence the niche differentiation of the two species when sympatric, as well as potentially explaining the differences in distribution in the two species.
There were species differences in some of the traits but not in others. Out of the traits that did differ between the two species, five were traits that dealt with water relations, and one was a trait related to the shapes and sizes of the leaves. The one trait related to leaf morphology was leaf ratio, which was a trait defined by the width and length of the leaf. We found a lower mean leaf ratio score in Silene dioica, as compared to Silene latifolia, meaning that S. dioica had wider leaves. S. dioica does have more broad and ovate leaves than S. latifolia (Baker, 1947), and the longer lanceolate leaves of S. latifolia likely contributed to their higher scores for the trait leaf ratio. Leaf morphology has many consequences for how plants utilize resources from their surroundings and deal with everything from the light environment
(McIntyre & Strauss, 2013; Sessa & Givnish, 2013) to stomatal conductance and water relations (Carlson et al., 2016; Taiz, 2014). Favre and Karrenberg (2011) have investigated the trait SLA in this species pair and how it responds to drought and shade stress and it was found that the response of the trait to these stresses was more phenotypically plastic in S. latifolia than in S. dioica. While I did not detect
significant differences in SLA between the species, this could have been because I was measuring very young leaves where the SLA differences are not detectable between the two species. The overall values captured in terms of SLA for both species were higher than those in another paper comparing the two species (Favre &
Karrenberg, 2011). This could reflect the fact that the plants measured in this study were in a greenhouse, while the ones in their paper were grown outside and were older plants. SLA has been shown to vary across large scale environmental gradients
(Swenson et al., 2012), and I expected this trait could reflect the different usage of water as it provides protection from damage due to dessication but there was no such significant variation found in the results.
The other traits that were different between species were ones related to water usage, such as turgid weight, fresh weight in the greenhouse at the time of picking and succulence. The fact that both species differ from each other in traits related to water usage, with S. latifolia able to absorb a greater maximum quantity of water in its leaf than that of S. dioica (turgidw), and that a trait related to uptake of water (wgain) is significantly associated to the climate variable across the range of S. latifolia points to a possible ecologically relevant difference between the two species. Divergence in these traits related to water usage may have contributed to S. latifolia’s propensity towards drier conditions, and thus resulted in its later reproductive isolation and speciation from S. dioica. It is not surprising I do not see this differentiation to the same extent on the population level within S. latifolia, as the populations we were looking at have likely been separated for a much shorter length of time and occupy more similar sites than the two species have been.
I detected species differences for many of the traits measured as discussed above, but there was also differentiation on the population level for many of the same traits. In traits where this differentiation was not present, the results may reflect homogenizing gene flow between populations of the same species, or the fact that the sites themselves were more similar to each other than when sites of S. latifolia was compared with S. dioica. One of the traits that differed with population for both S.
dioica and S. latifolia was the trait succulence. Populations at lower latitudes in S.
latifolia (ESP, LIB) had a lower mean succulence value, which went against what we expected to find as succulence is often linked with drought resistance (Azuma et al., 2016; Ogburn & Edwards, 2010; Razzaghi et al., 2015). There are precedents in the botanical world for succulence conferring frost resistance among members of Senecio and Lobelia by increasing the amount of cellular sucrose which acts as a protection from freezing for cell membranes (Beck et al., 1982; Steponkus et al., 1977), or by way of using the liquid in the leaves as a thermal buffer against cold damage (Chadwick, 1981), but it is not possible to theorize on these physiological
mechanisms in the case of Silene without further physiological investigations. It would be useful in future to attempt to disentangle more precisely the physiological differences that allow the different taxa to interact with water differently.
In terms of looking at how the results could be used to explain the differing distributions of the two species, I note that the occurrence of a species along an environmental gradient may be limited mainly by its ability to tolerate conditions of greater abiotic stress (Brown, 1996). In less stressful conditions, away from the upper ends of their range, biotic interactions such as competition may assume a greater role in explaining distribution (Normand, 2009). Water use efficiency is also an important climate related factor which impacts plant range (Walter et al., 2015) and can provide a constraint at species lower latitudinal range limit (Normand, 2009). The rate of water uptake (wgain) was a trait significantly related to the climate in Silene latifolia, and this as well as the differences in traits related to drought tolerance (succulence, turgidw) between the two species could show that water usage and drought tolerance plays a role in defining the lower boundary of each species’ range. The sites from the southern end of S. latifolia’s range (ESP1, LIB1) were more arid and had longer periods of drought with higher temperatures than any sites in S. dioica’s range, this could indicate that increased drought tolerance may have allowed the southern range of the species to expand past that which S. dioica could tolerate. The population at the lowest latitude for S. dioica was statistically different from the other populations in terms of wgain, and the three populations at the lowest latitude in S. latifolia had significantly different succulence levels than plants from other S. latifolia more northerly populations. This shows traits of water usage in the plants differentiating populations, as well as the species. Future studies of the two plants would ideally incorporate some information about their capacity to withstand freezing damage and to tolerate cold generally which could provide some context to the varying strength of ecological factors determining the niche differentiation across the two plants’ ranges.
Silene dioica in its northern range fulfils an additional different ecological role than in its more southerly ranges, acting as a roadside weed on waste ground particularly in eastern and northern Finland (Hathaway et al., 2009), while this niche is more commonly filled by S. latifolia elsewhere (Baker, 1947b). The fact that this northern part of S. dioica’s range is not one where S. latifolia is present may allow S. dioica to inhabit niches which are occupied across other parts of its range. Silene latifolia’s lack of tolerance for long winters and short growing seasons in comparison to S.
dioica’s (Favre et al., 2016) has been indicated as an example of differential adaptation between the two species. In future, traits that are typically important for
be useful metrics to investigate to further investigate the physiological basis of this differential adaptation. As the most reliable variation present in plant species over a large latitudinal gradient occurs during the period of reproductive development (Edwards et al., 2011; Griffith & Watson, 2005), it would additionally be useful to include measurements of traits of mature plants such as time to flowering or number of flowers. This is relevant because of the different floral syndromes expressed between the two species (Goulson, 1997; Waelti et al., 2008), and the indication that S. dioica prioritizes survival over flowering while S. latifolia does the opposite (Favre et al., 2016).
A follow up study would ideally also pursue a thorough investigation of the additive genetic component of quantitative traits (QST) related to the water gain trait as well as the other significant traits. The variation within each population was too high for this study to procure a QST value, which could relate to its sample size. A future study would ideally use a larger data set to minimize this within population variation. This future study could compare the QST values with neutral molecular markers to see if these traits represent adaptive differences between the species or populations, although caution must be used in concluding that selection is occurring from these comparisons (Miller et al., 2008). That being said, the QST/FST genetic distance values could allow future studies to indicate whether or not the variation present is due to selective processes or simply neutral processes such as genetic drift (Leinonen et al., 2013).
In terms of what the data from this study can tell us, it has potentially identified a number of candidate traits which are important for maintenance of the species boundary, this study indicates that there is limited evidence that there are some population differences within each species, and that traits relating to water relations may be potentially important for separating the two species, as well as within S. dioica. Further research will shed more light on whether these traits are adaptive and reflect local adaptation in these species, especially when paired with genetic information about neutral molecular markers.
Acknowledgments
First and foremost, I want to thank my supportive and kind supervisor Dr. Sophie Karrenberg. She provided expert advice and was very closely involved in every single step of the project and without her help and support it would never have been possible. I couldn’t have asked for a better supervisor. I would also like to thank everyone who has helped me over the duration of this thesis, including my family, Þórdís, and everyone at the EBC Research Facilities and the Greenhouse staff who helped make this project possible.
I would also like to dedicate this thesis to my father.
