Figure captions:
Figure 1 . Examples of stress induced behavioural shifts. A)
Mean values of dive duration and distance travelled of individuals of
ringed seal (Pusa hispida ) in Svalbard (Norway) before and after
a significant decline in sea-ice extension, and thus in their habitat
quality. Both behaviours showed significant increases in response to the
stress. Bars indicate standard error. Data adapted from Hamilton et al.
(2015). B) Probability of aggressive encounters between heterospecific
and conspecific butterflyfishes (Chetodon spp.) calculated in 4
regions across the central Indo-Pacific, before and after a mass coral
bleaching event in 2006 (reduction in food resources for the fishes).
Both behaviours showed significant decreases. Data adapted from Keith et
al. (2018). Bars are 95% confidence intervals. *Coral cover % change
showed refers only to one of the four sites (Christmas Island site,
Indian Ocean; 105.6° E, 10.4° S) for simplicity.
Figure 2 . Examples of stress induced shifts in morphological
traits. A) Mean estimated body mass of female polar bear individuals
(Ursus maritimus ) in western Hudson Bay from 1980 through 2007
(dashed line indicates fit of linear regression [r=-0.549,
p<0.01], bars indicate standard error); the area experienced
progressively earlier dates of sea-ice breakup, that is a decline in
duration of the favorite sea-ice habitat of polar bears. Data adapted
from Stirling & Derocher 2012. B) Comparison of the mean of individual
asymmetry index for femoral pores among Italian wall lizard
(Podarcis siculus ) sampled in hazelnut orchards with no history
of pesticide use (control) vs orchards regularly treated with pesticides
(treatment). The pollution stress induced an increase in the fluctuating
asymmetry for femoral pores (bars indicate standard error). Data adapted
from Simbula et al. 2021.
Figure 3. Theoretical example of a timeline to collapse. Here
we posit a population of seabirds inhabiting an area where prey
resources (e.g. fish stocks) begin a continuous decline (A). The curves
in B and C represent average values of a behavioural (B) and
morphological trait (C) calculated from a pool of individuals in the
population through time. The red curve in D shows the abundance of the
population. First a shift is observed in the behaviour (time point
TBs), where the average foraging distance increases
compared to the average measured during stable conditions
Bs (B). The foraging distance will increase until it
reaches a physiological limit (time point TBe), defining
the time interval where a continuous change is observable
(IB). After, or during such time, we will observe a
decrease in average body size compared to that measured during stable
conditions Ms (C), at time TMs. The body
size will change until its physiological limit (TMe),
defining the time interval where such continuous change is observable
(IM). Later, the abundance trend of population will show
alterations in the pre-decline indicators such as Early Warning Signals
(EWSs), that will start to be observable at time point
TAs, and will last until
TAe, defining the
time interval IA. Subsequently, the continuous decreases
to extinction (D) will begin at time point TEs, and end
will end with the extinction of the population TEe,
lasting the time interval IE. The first occurrence of
the signals projected on the lower Time axes shows the sequence in the
category of observable signals of stress starting at the individuals’
level (B, C) and propagating to the population level (D).
Figure 4. Theoretical example of a timeline to collapse. The
grey curve represents a continuous growth of a given environmental
stress on the population. The blue and green curves represent average
values of a behavioural and morphological trait calculated from a pool
of individuals in the population. The red curve represents the number of
individuals. Bs and Ms are respectively
the average measure of the considered behavioural and morphological
traits in stable conditions. The small black dotted lines project the
starting point of the shifts in morphological traits and abundance
dynamics on the behaviour (point 1) and morphological trait (point 2)
curves. Projected on the vertical axis, those points identify
Bx and Mx: the values of behavioural and
morphological metrics at the time of the onset of the next signal along
the timeline. The interval of change (brackets) from the average values
defines the intrinsic buffering capacities of behaviour
(Cb,) and morphological traits (Cm).
Figures.