Figure legends
Figure 1. Predictability of trait genetic differentiation among
populations after comparison of experimental data and observational
field data. In a common garden experiment with individuals from multiple
provenances (“source” environments) growing in a set of treatments
(“exposure” environments), one can partition the independent (a-e) or
interacting (f-h) effects of genetic differentiation and phenotypic
plasticity. Population genetic differentiation is identified as trait
variation along source environments (blue lines), and plasticity is
detected by comparing trait values between low (light blue) and high
(dark blue) levels of the exposure environment. Note that source and
exposure environments are driven here by the same underlying
environmental factor. The resulting pattern expected to be observed
across field populations is shown with red dashed lines, linking two
extreme populations along the environmental gradient in the treatment
closest to their corresponding source conditions (from low to high
treatments). Observational field data will provide a reliable prediction
of genetic differentiation (green squares) in the presence of (a) only
source effects, (c) source and exposure effects with the same direction,
or (f) source effects with consistent direction but inconsistent slope
across exposure environments. In contrast, genetic differentiation will
not be predictable from field patterns (black squares) in the presence
of (b) only exposure effects, (d,e) source and exposure effects with
opposite directions (“countergradient variation”), or (g,h) source
effects with inconsistent direction across exposure environments. Note
that the x-axis contains source rather than exposure environment (this
differs thus from classical displays of plasticity in reaction norms, as
the exposure environment is best envisaged as a set of treatments with
observations over a continuous environmental gradient comprising the
source environments).
Figure 2. Location of native (black) and non-native (grey)
study populations of Plantago lanceolata in geographical (a) and
environmental (b) space. Circles indicate populations studied in the
field; triangles indicate populations studied in the field and included
in the greenhouse experiment. Colours filling the world map in a)
correspond to mean annual temperature and precipitation as shown in b).
In b), small black and grey background points correspond to the
environmental niche occupied by the species in the native and non-native
range, respectively, according to occurrence data from GBIF and BIEN
databases.
Fig. 3. Effects from the best model (blue) and competing models
(grey; ∆AICc < 2) for each trait ofPlantago lanceolata in the greenhouse, with 95% confidence
intervals. The effects correspond to source environmental drivers (A =
Aridity; T = Temperature; C = Vegetation Cover; M = Mowing),
experimental treatments of Water (Wd = dry) and Light
(L64 and L33), and the interactions
between them. Vegetative traits (a-c) are biomass, specific leaf area
(SLA) and root:shoot ratio (RSR), and reproductive traits (d-e) are
probability of flowering (“Flw Prob”) and fecundity (“Fecund”). For
simplicity, we omit the effects of control biomass. The effects of
L64 treatment and Mowing were not tested in RSR and
fecundity, respectively (absent labels; see Material and methodsfor details).
Figure 4. Effects of two source environmental drivers (A =
Aridity; C = Vegetation Cover) and their corresponding exposure
treatments (Water and Light) on Plantago lanceolata traits in the
greenhouse. Vegetative traits are biomass (a-c), specific leaf area
(SLA; d-f) and root:shoot ratio (RSR; g-i), and reproductive traits are
probability of flowering (Flw Prob; j-l) and fecundity (Fecund; m-o).
Results are presented with 95% confidence intervals (CI), and
correspond to the best model according to Akaike Information Criterion
(empty subpanels or bars indicate no effect in the best model). In the
left and middle columns, trait values are shown for wet
(Ww) and dry (Wd) water treatments, and
for L100, L64 and L33light treatments. All traits are mean centred and scaled by the standard
deviation, except for probability of flowering (Y-axis in logit scale).
Source drivers are mean centred and scaled by two times the standard
deviation (see Appendix S3). The distribution of populations along
source environment values is shown by rug marks on the inside of the x
axis. In the right column, the effects of source environment (genetic
differentiation; yellow), exposure environment (plasticity; orange) and
their interaction (red) are compared. Note that effect sizes are given
as absolute values for comparison, and only the CI upper limit is shown.
Fig. 5. Effects from the best model (blue) and competing models
(grey; ∆AICc < 2) for each trait ofPlantago lanceolata in the field, with 95% confidence intervals.
The effects correspond to environmental factors (A = Aridity; T =
Temperature; C = Vegetation Cover; M = Mowing), non-native range
(Rnnat), and the interactions between them. Vegetative
traits (a-b) are biomass and specific leaf area (SLA), and reproductive
traits (c-d) are probability of flowering (“Flw Prob”) and fecundity
(“Fecund”). The effects of environmental factors alone correspond to
native populations; the effects of environmental factors on non-native
populations can be deduced by summing environmental effects alone and
the effects of range × environment interactions. For simplicity, we omit
the effect of biomass in models of probability of flowering and
fecundity.