Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Research

OIKOS
Allee effects under climate change

Luděk Berec

L. Berec (http://orcid.org/0000-0002-2419-3324) ✉ (berec@entu.cas.cz), Centre for Mathematical Biology, Inst. of Mathematics, Faculty of Science, Univ.
of South Bohemia, Branišovská 1760, CZ-37005 České Budějovice, Czech Republic, and: Czech Academy of Sciences, Biology Centre, Inst. of Entomology,
Dept of Ecology, Branišoská 31, CZ-37005 České Budějovice, Czech Republic.

Oikos Understanding how climate change affects population dynamics is crucial for assessing
128: 972–983, 2019 future of biodiversity. Here I ask how can Allee effects, occurring when mean individ-
doi: 10.1111/oik.05941 ual fitness is reduced in rare populations, respond to increasing temperature. Despite
the role Allee effects play in ecology of invasive, threatened and harvested populations,
Subject Editor: Björn Rall impacts of climate change on Allee effects are practically unknown. Analysis of two
Editor-in-Chief: Dries Bonte population models reveals that whereas the Allee effect driven by predation gener-
Accepted 25 January 2019 ally weakens as temperature increases, the Allee effect due to need of finding mates is
predicted to become stronger when warming occurs. For the former model, the meta-
bolic theory suggests that with increasing temperature prey growth rate should increase
faster than predator attack rate. Increasing temperature thus weakens the Allee effect.
In the latter, gypsy moth population model, mating rate increases with warming due
to enhanced female−male encounter rate and temperature-induced modifications in
female and male adult emergence distributions. However, male and female mortality
rates increase, too and the net effect is strengthening of the Allee effect. These results
have repercussions also for pest control, indicating that augmentation of biocontrol
agents may perhaps be not as effective as using pesticides or disrupting mating.

Keywords: global warming, metabolic theory

Introduction
Understanding how climate change impacts population and community dynamics
is crucial for assessing potential changes in global biodiversity (Pereira et al. 2010,
Bellard et al. 2012). Many predictions are quite warning. These include reduced popu-
lation sizes or densities, elevated risk of population extinctions, enhanced frequency of
non-native species invasions, increased pest species activity in their native habitats and
shifts in geographical species ranges (Pereira et al. 2010, Bellard et al. 2012). On the
other hand, enhanced temperature variation (Berec et al. 2013, Lawson et al. 2015),
thermal acclimation (Sentis et al. 2015) and warming-induced changes to body size
(Osmond et al. 2017, Sentis et al. 2017) were all suggested to at least alleviate some of
the negative impacts of global warming.
A growing interest in impacts of global warming on community dynamics has
recently showed up, focusing on individual predator−prey interactions (Dell et al.
––––––––––––––––––––––––––––––––––––––––
© 2019 The Author. Oikos © 2019 Nordic Society Oikos
www.oikosjournal.org

972
2014, Fussmann et al. 2014, Gilbert et al. 2014, Sentis et al. (a) x 10−6
2015, Osmond et al. 2017, Uszko et al. 2017), tri-trophic 1
food chains (Binzer et al. 2012, Sentis et al. 2017) and even logistic−like growth
complex food webs (Binzer et al. 2016). In all these stud- weak Allee effect

Per capita growth rate g(BR)


strong Allee effect
ies, the basal species (i.e. the prey species at the lowest food 0.5 fatal Allee effect
web level) were assumed to follow logistic dynamics. None
of these studies thus covered the case of a positive relation-
ship between the per capita population growth rate and
population density, a phenomenon termed the Allee effect 0
and observed or at least presumed to occur in many plant
and animal populations (Stephens and Sutherland 1999,
Courchamp et al. 2008). −0.5
Allee effects are not just a marginal theoretical construct,
but a vital process affecting many important and applied
aspects of population dynamics. These include chance of
invading populations to establish and spread (Tobin et al. −1
0 1 2 3 4 5
2011, Vercken et al. 2011), risk of small and threatened popu- Prey biomass−density BR
lations to go extinct (Brashares et al. 2010, Crates et al. 2017)
and danger of economically important populations of getting
(b) x 10−6
overharvested (Hutchings 2015, Haider et al. 2017). In all 1
these areas of applied ecology, acquaintance of presence and BC = 0.025
strength of Allee effects is crucial (Stephens and Sutherland Per capita growth rate g(BR) BC = 0.045
1999, Courchamp et al. 2008). Understanding whether Allee
effects can (dis)appear due to climate change and how they 0.5
can respond to it is thus of paramount importance for both
basic and applied ecology.
Allee effects occur when the per capita population growth 0
rate declines with decreasing density in rare populations, and
are classified as weak, strong or fatal (Fig. 1a; Stephens et al.
1999, Boukal and Berec 2002, Courchamp et al. 2008).
−0.5
Populations with weak Allee effects have positive per capita
growth rates at any density. If an Allee effect is strong, a criti-
cal population density, termed the Allee threshold, exists such
that the population practically declines to extinction when −1
0 1 2 3 4 5
reduced below the Allee threshold. Finally, fatal Allee effects
occur when the per capita growth rate is negative at any pop- Prey biomass−density BR
ulation density. Since fitness components of many species are Figure 1. Per-unit-biomass prey growth rate g(BR) (7) under an Allee
influenced by temperature (ectotherms comprise over 99% effect. (a) Possible types of Allee effect: weak (black solid,
of all species; Atkinson and Sibly 1997), the type of Allee BC = 0.01 g m−2), strong (black dashed, BC = 0.03 g m−2) and fatal
effect (weak, strong or fatal) and the Allee threshold in case of (black dash-dot, BC = 0.06 g m−2). Logistic-like growth occurs if
strong Allee effects may be temperature-dependent and hence predator biomass-density is low enough (grey, BC = 0.002 g m−2). (b)
respond to warming. Strong Allee effect: the black dots are the respective Allee thresholds
Intriguingly, impacts of increasing temperature on the and the gray lines are the respective tangent lines to g(BR) at these
strength and more fundamentally on a possibility of (dis) thresholds. The slope of these tangent lines is the Allee slope. Other
appearance of Allee effects remain virtually unexplored. parameters as in Table 1; mR = 10−4 g, mC = 10−3 g, T = 15 K.
Current knowledge is limited to a modeling study of the
invasive spiny water flea Bythotrephes longimanus subject and thus comprises different mechanisms to be affected by
to enhanced mating difficulties at lower densities (a source temperature.
of Allee effect; Gascoigne et al. 2009), in which the Allee
threshold was shown to decline with increasing water tem- Allee effects driven by predation
perature (Wittmann et al. 2011). Here I explore how could
Allee effects respond to climate change. Specifically, model- Methods
ing is used to examine how could the type of Allee effect
(weak, strong or fatal) and the Allee threshold in case of A predation-driven Allee effect occurs in the prey
strong Allee effects vary with temperature. In any of the fol- population when individual prey are more vulnerable to
lowing two sections I present a population model and its predation as prey density declines (Gascoigne and Lipcius
analysis. Each model has a different source of Allee effect 2004, Courchamp et al. 2008, Kramer and Drake 2010).

973
This happens, for example, when a logistically growing body masses and are commonly factored as (Kalinkat et al.
prey (or resource) population of a biomass-density BR is 2013, Binzer et al. 2016, Sentis et al. 2017)
attacked by a predator (or consumer) population with a
Holling type II (i.e. decelerating) functional response
and a constant biomass-density BC (Gascoigne and R C  T −T 
a = a0mRsa mCsa exp  − E a 0 (4)
Lipcius 2004, Courchamp et al. 2008, Berec et al. 2018).  kTT0 
Denoting by r and K the maximum per-unit-biomass prey
growth rate and the prey carrying capacity (in terms of
biomass), respectively, by a the predator attack rate and R C  T −T 
by h the predator handling time, the corresponding prey h = h0mRsh mCsh exp  − E h 0 (5)
population model is  kTT0 

Here mC is the predator body mass, while the other param-


dBR  B  aBR eters have a meaning analogous to the respective ones for r
= rBR 1 − R  − BC (1)
and K. The specific values used to parameterize Eq. 2–5 are
dt  K  1 + ahBR
summarized in Table 1.
The assumption of constant predator biomass-density To explore impacts of warming on the type and strength
BC is a reasonable approximation e.g. when the preda- of Allee effect, the temperature T is varied. I vary also
tor’s assimilation efficiency is low and/or the predator is the predator biomass-density BC, which will allow one to
a generalist with dynamics largely dependent on other, speculate on what happens if BC itself is subject to climate
unstudied prey species (Gascoigne and Lipcius 2004, change. Indeed, the predator biomass-density BC, due to
Berec et al. 2018). With model 1, I examine how the type its link to dynamics of the predator’s primary prey, may
and strength of Allee effect vary with temperature. I note increase or decrease with warming (Vasseur and McCann
that model 1 is formally equivalent to the basal species 2005). For each combination of T and BC, I determine
(i.e. the prey species at the lowest food web level) equa- the type of Allee effect (weak, strong or fatal) and the prey
tion of many recent models exploring impacts of climate biomass-density at positive stable equilibrium. If a strong
change on community dynamics (Binzer et al. 2012, 2016, Allee effect occurs I further calculate the Allee threshold
Dell et al. 2014, Fussmann et al. 2014, Gilbert et al. 2014, and the Allee slope. The concept of Allee slope, recently
Sentis et al. 2015, 2017, Osmond et al. 2017, Uszko et al. introduced by Walter et al. (2017) and defined as a mag-
2017). nitude of positive relationship between the per capita (here
Interpreting model 1 in terms of biomass-densities per-unit-biomass) population growth rate and density (here
instead of densities of individuals makes possible to use biomass-density) in the vicinity of the Allee threshold,
the metabolic theory of ecology and thus express relevant has important implications for population dynamics and
model parameters in terms of body mass and temperature persistence (Walter et al. 2017).
(Brown et al. 2004). Body mass and temperature depen- For the parameter values specified in Table 1, the prey
dence of the maximum per-unit-biomass prey growth rate carrying capacity K declines as the temperature T increases.
r is commonly formalized as (Binzer et al. 2012, 2016, However, contrary to the prediction of the metabolic theory,
Sentis et al. 2017) some studies reported temperature-independent carrying
 T −T  capacity (Enquist and Niklas 2001) or even its positive rela-
r = r0mRsr exp  − Er 0 (2) tionship with temperature (DeLong and Hanson 2011); see
 kTT0 
Table 1. Parameter values for the body mass and temperature-
where r0 is the normalization constant independent of body dependent relationships (Eq. 2–5). Values for the maximum per-unit-
mass and temperature, mR is the prey body mass, sr is the biomass growth rate r (s−1) are from Savage et al. (2004) and
allometric exponent, Er is the activation energy (eV), k is the Sentis et al. (2017), those for the carrying capacity K (g m−2) are from
Boltzmann’s constant (8.62 × 10−5 eV K−1), T is the absolute Meehan (2006) and Fussmann et al. (2014), and those for the attack
rate a (m2 s−1) and the handling time h (s) are from Binzer et al.
temperature (K) and T0 is the normalization temperature (K; (2016) and Sentis et al. (2017). Body masses are considered in grams
here 20°C = 293.15 K). Similarly, the prey carrying capacity (g). The last line indicates whether the respective parameters increase
K can be expressed as (Binzer et al. 2012, 2016, Sentis et al. (↑) or decrease (↓) with increasing temperature T.
2017)
Parameter (x) r K a h
 T −T  Unit s−1
gm −2 2 −1
m s s
K = K 0mRsK exp  − E K 0 (3)
 kTT  0
x0 1.55 × 10−7
−0.25
31.19
0.28
2.04 × 10−6
0.25
15.68 × 103
−0.45
s x(R )
where K0 is the normalization constant independent of body × × −0.8 0.47
mass and temperature, sK is the allometric exponent and EK s Cx
is the respective activation energy. The predation-related Ex (eV) 0.84 −0.71 0.38 −0.26
parameters a and h of model 1 feature both prey and predator Increase in T ↑ ↓ ↑ ↓

974
also discussions in Vasseur and McCann (2005), Dell et al. the actual predator biomass-density BC, Allee effects change
(2014) and Gilbert et al. (2014). Moreover, K is not a model from fatal to strong, from strong to weak, or from weak
parameter linked to a specific process, but rather a complex to a logistic-like prey growth (Fig. 2a). The temperature at
outcome of population dynamics. Therefore, I examine also which strong Allee effects become weak corresponds to when
impacts of neutral or positive relationships between the g(0) = r − aBC = 0 or equivalently when BC = r/a. As tempera-
temperature T and the prey carrying capacity K. In addi- ture grows, both the prey growth rate r and the predator
tion, I study a version of model 1 in which the prey popula- attack rate a increase, yet r increases at a higher rate than
tion grows exponentially rather than in a logistic manner. a (Er > Ea in Table 1) and so Allee effects weaken with T.
This is not a problem in the context of this study, since Similarly, the temperature at which fatal Allee effects become
Allee effects affect population dynamics predominantly at strong corresponds to when the expression under the radical
low (biomass-)densities where within-species competition is of Eq. 6 equals zero or equivalently when BC = r(1 + ahK)2/
likely comparably weak. (4a2hK). Although we cannot infer much from this formula,
Fig. 2a suggests that this boundary can be parabolic. Indeed,
Results for values of BC around 0.017 Allee effects can change with
warming first from strong to fatal and later from fatal back
Model 1 has up to two positive equilibria. Assuming BC > 0 to strong (Fig. 2a). Still, even though here an increase in
(otherwise BR = K is the only positive equilibrium of prey), temperature may rather worsen things and cause population
these can be obtained by setting the right-hand side of model extinction, the general pattern appears to be that increasing
1 to zero and solving the resulting equation for the prey temperature alleviates any Allee effect present.
biomass-density BR: For strong Allee effects, Allee thresholds are smaller at
higher temperatures (Fig. 2b). Moreover, Allee slopes are
steeper at higher temperatures (Fig. 2c). Smaller Allee thresh-
( K −1 / (ah )) ± ( K −1 / (ah )) ( )
2
+ 4 K / ( ah ) 1 − ( a / r ) BC
BR1,2 = (6) olds are thus more clearly distinguished: prey populations
2 with biomass-densities just above those thresholds grow
With the per-unit-biomass prey growth rate denoted as faster and hence escape from staying close to them more
easily (Fig. 1b). On the contrary, larger Allee thresholds are
1 dBR  B  aBC more blurred: prey populations with biomass-densities just
g ( BR ) = × = r 1 − R  − (7)
BR dt  K  1 + ahBR above those thresholds grow slowly and are thus vulnerable
to extinction for a prolonged period of time, especially when
an Allee effect occurs if g(BR) increases close to the zero prey subject to repeated perturbations (Fig. 1b; Meyer 2016). The
biomass-density BR = 0, that is, if g′(0) = −r/K + BCa2h > 0 prey populations subject to Allee effects driven by predation
or equivalently BC > r/(a2hK). Otherwise, the prey popu- are less vulnerable to extinction at higher temperatures also
lation growth is logistic-like, with g′(BR) < 0 at any BR ≥ 0. from this perspective.
An Allee effect is weak (one stable positive equilibrium) While the Allee threshold declines with increasing tem-
if g(0) = r − aBC > 0 or equivalently BC < r/a, strong (two perature (Fig. 2b), the positive stable equilibrium decreases,
positive equilibria of which the higher is stable and the too (Fig. 2d). An important practical question then is how
lower is unstable) if g(0) < 0 or equivalently BC > r/a and the their difference changes with temperature. This difference is
expression under the radical of Eq. 6 is positive, and fatal a measure of prey population resilience to perturbations in
(no positive equilibrium) if g(0) < 0 or equivalently BC > r/a its biomass-density, as it represents the minimum magnitude
and the expression under the radical of Eq. 6 is negative of single perturbation required to get prey below the Allee
(Fig. 1a). In case a strong Allee effect occurs, the lower, unsta- threshold, right to the basin of attraction of the stable extinc-
ble equilibrium is the Allee threshold and the slope of the tion equilibrium (Meyer 2016). Interestingly, this (common)
tangent line to g(BR) at the Allee threshold is the Allee slope measure of population resilience declines with warming
(Fig. 1b; Walter et al. 2017). (Fig. 3).
It is instructive to first look at how changes in any single As discussed above, the prey carrying capacity K may be
parameter in model 1 affect the type and strength of any unaffected by temperature or even increase with warming.
potential Allee effect. With the expressions from the previ- Since increasing K decreases chances of an Allee effect to occur
ous paragraph it is not difficult to see that when the maxi- and increases the critical predator biomass-density above
mum per-unit-biomass prey growth rate r decreases or when which strong Allee effects become fatal, these alternative
the predator attack rate a or biomass-density BC increase, the relationships between the prey carrying capacity and temper-
chance of an Allee effect to occur increases and an Allee effect ature only corroborate the general prediction that warming
shifts from weak to strong and eventually to fatal (Fig. 1a). weakens Allee effects driven by predation (Supplementary
Similarly, increasing the prey carrying capacity K or the preda- material Appendix 1 Fig. A1, A2). Moreover, under both
tor handling time h raises the chance of an Allee effect to occur. temperature-independent and positively temperature-
For the parameters specified in Table 1, Allee effects gen- dependent prey carrying capacity, the prey biomass-density
erally weaken as the temperature T increases. Depending on at positive stable equilibrium increases with temperature

975
Figure 2. Dependence on the temperature T and the predator biomass-density BC of the (a) Allee effect type, (b) Allee threshold, (c) Allee
slope and (d) prey positive stable equilibrium. Parameters as in Table 1; mR = 10−4 g, mC = 10−3 g. Color legend in panel (a): white = logistic-
like growth, red = weak Allee effect, green = strong Allee effect, blue = fatal Allee effect. Color legend in panels (b–d): the respective variable
increases in the direction from dark blue to dark red.

and so does population resilience (Supplementary material effect in its prey (Kot 2001). Also, this situation has been
Appendix 1 Fig. A1, A2). Last but not least, the results of the studied repeatedly (Vasseur and McCann 2005, Dell et al.
model with exponential prey growth are comparable to those 2014, Gilbert et al. 2014, Osmond et al. 2017, Uszko et al.
of the models in which the prey population grows in a logistic 2017). These studies reveal, among other things, that a nega-
manner (Supplementary material Appendix 1 Fig. A3). These tive relationship between temperature and body size may
results also suggest a potential mechanisms behind a non- reduce impacts of warming on consumer−resource dynamics
monotonic boundary between strong and fatal Allee effects (Osmond et al. 2017), that increasing temperature may cause
observed in Fig. 2a. Comparison of this figure with Fig. a transition in predator−prey dynamics from coexistence
A1a, A2a indicates that low negative values of the activation at a stable equilibrium to sustained predator−prey cycling
energy EK of the prey carrying capacity may govern this non- (Vasseur and McCann 2005), and that predator biomass-
monotonicity: in the latter two figures one cannot get a fatal density may initially increase with warming and later decrease
Allee effects from a strong one just by increasing temperature. (Vasseur and McCann 2005). The latter two predictions are
Model 1 does not cover a situation where the predator especially important from our perspective, since the preda-
is a specialist and through a numerical response is dynami- tor biomass-density BC may respond to warming through the
cally linked to its (primary) prey. The major reason for not interaction with its (unmodeled) primary prey. One can envi-
studying this situation is that even with a Holling type II sion drawing an anticipated predator trajectory in the T − BC
functional response such predator does not trigger an Allee plots like in Fig. 2 or Supplementary material Appendix 1

976
0.03 males or females as Gaussian distributions with means (i.e.

Predator biomass−density BC
Resilience dates of peak emergence) µM and µF (and with µF > µM due to
0.025 protandry) and SD σM and σF for males and females, respec-
tively (Robinet et al. 2007). Since the exact date of peak male
0.02
emergence (µM) by itself does not affect an overlap between
0.015 male and female maturation distributions and hence the
strength of Allee effect, µM is set arbitrarily to 5σM so that
0.01 virtually no males emerge before day 0 (Robinet et al. 2008,
Walter et al. 2015). Mating dynamics are then described by
0.005 a continuous-time mating model. Denoted by R, the mat-
ing model outputs the number of females Ft m that success-
0 fully reproduce by the end of mating season. Each of the
0 10 20 30
Temperature T
mated females lays one egg mass composed of λ eggs, time
is enhanced by one (generation) and the just described life
Figure 3. Dependence on the temperature T and the predator bio- cycle is repeated. This gives the following model of between-
mass-density BC of the prey population resilience, defined here as generation dynamics:
the difference between the prey biomass-density at the positive sta-
ble equilibrium and the Allee threshold, under strong Allee effects.
Parameters as in Table 1; mR = 10−4 g, mC = 10−3 g. Color legend as in
( ( )
J t +1 = λFt m = λR J t a 1 − p W , µ M , µ F , σ M , σ F ) (8)
Fig. 2. I now describe the continuous-time mating model R. Whereas
females of the gypsy moth mate only once, males may mate
Fig. A1–A3 and see what happens to the Allee effect type and multiply. The mating model thus considers five moth states,
strength. For example, cycling dynamics of BC would mean each expressed in terms of density (numbers per hectare):
that the strength of Allee effect and even its type may fluctu- virgin females (V), fertile males searching for females (Ms),
ate. Alternatively, a possible hump-shaped form of the preda- virgin female–fertile male couples (Q), fertile males tempo-
tor biomass-density BC as a function of temperature would rarily resting after mating (Mr) and mated females (Fr). Virgin
mean strengthening Allee effects initially and their weakening females and fertile males encounter at the rate z and have the
later on. In any case, the presented results allow for a variety probability c that courting is successful. If this is the case,
of predictions of this kind. tq is the mean time spent in the couple. Males need to rest
after each mating; let their mean resting time be tr. Within
Allee effects due to need of finding mates the mating season, males and females die at rates dM and dF,
respectively. Since mated females Fr are supposed to oviposit
Methods shortly after mating, they are assumed not exposed to mor-
tality. These assumptions leave the following mating season
A mate-finding Allee effect occurs when individual females model (Blackwood et al. 2012):
have enhanced difficulty to locate (compatible) mates
as population density declines (Courchamp et al. 2008, dV
Gascoigne et al. 2009, Shaw et al. 2018). This Allee effect has
dt
( )
= 0.5 J t a 1 − p WN (µ F , σ F ) − zVM s
been demonstrated, for example, in the gypsy moth Lymantria
1
dispar, a pest of North American forests (Tobin et al. 2009). In
this species, males mature and emerge before females, a phe-
(
+ d M Q − d FV + 1 − c Q
tq
)
nomenon termed protandry (Robinet et al. 2007). Protandry dM s
affects the degree to which males and females are tempo-
dt
( )
= 0.5 J t a 1 − p WN ( mM , sM ) − zVM s
rally segregated and hence affects the strength of Allee effect
1 1
(Robinet et al. 2008, Blackwood et al. 2012, Walter et al.
2015). In the following paragraphs, I review the gypsy moth
(
+ d FQ − d M M s + 1 − c Q + Mr
tq tr
) (9)
population model due to Blackwood et al. (2012) and mod- dQ 1
ify it by introducing temperature dependencies in it. = zVM s − d M Q − d F Q − Q
The gypsy moth is a species with non-overlapping genera- dt tq
tions. Within a generation starting at time t, a proportion a dM r 1 1
of Jt eggs survive through the larval to pupal stage after under- = −d M M r + c Q − M r
dt tq tr
going background mortality. Pesticides may be applied at the
larval stage and only a proportion 1 − p of larvae then survive. dFr 1
=c Q
A fraction W of pupae then avoid predation by rodents so dt tq
that eventually Jta(1 − p)W pupae emerge. There is a 1:1 sex
ratio at birth. Adults do not emerge all at once. One may The first term in the equation for virgin females V says that
envision temporal distributions of the dates of emergence of there are 0.5 Jta(1–p)W females ready to mature and emerge

977
during the mating season and that the individual emergence temperature is 19.89°C (<www.usclimatedata.com/cli-
dates follow a Gaussian distribution with mean (the peak mate/logan/west-virginia/united-states/uswv0442>). As far
date) µF and SD (maturation dispersion) σF. The first term as I know, protogynous gypsy moth populations for which
in the equation for fertile searching males Ms has an analo- µF < µM do not exist. Still, setting elevation and northing
gous meaning. The second term zVMs in the equations for to zero in Eq. 10 gives µF < µM. However, the relationships
V and Ms represents the number of encounters between vir- derived by Walter et al. (2015) are statistical fits and as such
gin females and searching males and hence pairs formed per have questionable validity for extrapolating to areas outside
unit time. It is this rate which, for any fixed temperature, is of the gypsy moth range in North America. Moreover, the
responsible for occurrence of the mate-finding Allee effect in values of protandry generated by these fits for model 9 never
the gypsy moth population (Berec 2018, Shaw et al. 2018). became negative.
Since no adults are present in the population at the begin- Furthermore, I assume a constant temperature lapse rate
ning of mating season, all five state variables of the mating (i.e. decrease of temperature with altitude) of 6.5°C km−1,
model 9 are initially set to zero. It is the inflow of emerging the commonly used overall average rate (Minder et al. 2010).
adults that makes these variables positive as time proceeds. Hence,
The number of successfully mated females at the end of mat-
ing season is Ft m = Fr (t ) . Here τ is the length of mating
season, assumed to end 5σF days after µF so that virtually (
T = T (Logan ) + 0.0065 × elevation (Logan ) − elevation ) (11)
no females emerge beyond that day (Robinet et al. 2008,
Walter et al. 2015). All parameters involved in model 9 are From Eq. 11 elevation can be calculated as a function of
listed in Table 2. temperature and plugged in Eq. 10. Since I am interested
I now make some parameters of the models 8 and 9 here just in trends and conceptual understanding of impacts
temperature-dependent. Walter et al. (2015) showed that of climate change on Allee effects, precise numbers are less
both the degree of protandry (i.e. µF − µM) and maturation important, especially since different values produce analogous
dispersion (i.e. σM and σF, assumed equal) increased with results.
northing or elevation of the population’s location: Analogously to the parameters of model 1, the back-
ground mortality rates dM and dF can be made dependent on
µ F − µ M = −3.639 × 10−3 + 1.112 × 10−3 elevation body size and temperature:
+ 8.401 × 10−7 northing
(10)  T −T 
σ M = σ F = 23.09 + 1.172 exp (elevation/1000) d X = d 0X mXsd exp  − E d 0 (12)
−5 −12  kTT0 
− 1.358 × 10 northing + 2.067 × 10 northing 2

where X = M or F, sd = −0.31 and Ed = 0.69 eV (Sentis et al.


Both elevation and northing are measured in meters. Since 2017); both dM and dF thus increase with warming. Assuming
higher northing or elevation generally mean lower (mean) that the male and female body masses are mM = 0.5 g and
temperature, both these quantities tend to decrease with mF = 2 g (values that lie within the range of pupal weights,
increasing (mean) temperature. I arbitrarily choose Logan, Patrick C. Tobin, pers. comm.; intriguingly, adult body
WV, USA, as a reference location in the middle of gypsy moth masses do not appear to have been reported in the literature
infestation area, for which northing is about 4 188 839 meters, so I use the pupal body weights as a proxy), I estimate the
elevation is about 207 meters and (yearly average maximum) constants d 0M and d 0F so that dM = 1.67 and dF = 0.73 at

Table 2. Parameters of the gypsy moth model (Eq. 8, 9). After Blackwood et al. (2012).

Parameter Value(s) Definition Unit


λ 546 number of eggs per egg mass –
a 0.05 fraction of eggs surviving to pupation –
p Varies fraction of larvae killed by pesticides –
W 0.84 fraction of pupae surviving rodent predation –
µM 5σM peak date of male emergence day
σM Varies standard deviation of male emergence day
µF Varies peak date of female emergence day
σF σM standard deviation of female emergence day
z 0.15 encounter rate between females and males day−1
tr 1 mean resting time for males after mating day
tq 0.0042 mean time spent per day in a couple (6 min) day
c 0.91 probability of successful courting day
dM 1.67 background mortality rate for males day−1
dF 0.73 background mortality rate for females day−1
τ µF + 5σF mating season length day

978
T = T0 = 19.89°C as in Table 2, giving d 0M = 1.36 and Results
d 0F = 0.91 .
Activity and movement rate are also generally functions of Also here, it is instructive to first look at how changes in
temperature (Dell et al. 2014). Therefore, the encounter rate any single model parameter impact the Allee effect strength
z between virgin females and searching males likely depends or the Allee threshold value. Obviously, by increasing λ, a
on temperature, too. To quantify this dependence, one may and W or by decreasing p the Allee effect weakens and the
use a relationship originally derived for the consumer attack Allee threshold declines. Moreover, by increasing the degree
rate of the resource (Osmond et al. 2017). Assuming non- of protandry µF − µM and background mortalities dM and dF
flying virgin females (as is the case of gypsy moths in the or by decreasing the encounter rate z between virgin females
North American forests) and velocity v for searching males, and searching males the Allee threshold becomes larger.
the encounter rate z is Maturation dispersion has a non-uniform effect (Fig. 4a).
With increasing σM = σF, any effect of protandry gets reduced
as the male and female distributions of the emergence date
 T −T  increasingly overlap. This initially enhances the mating suc-
z = z0v0mMsv exp  − Ev 0 (13)
 kTT0  cess and hence decreases the Allee threshold. However, when
σM = σF become large enough, the distributions become quite
flat and the mating success low throughout the whole mat-
where sv = 1 and Ev = 0.46 eV (Osmond et al. 2017). Again, I ing season. Finally, by increasing c and by decreasing tq or tr
estimate the product z0v0 so that z = 0.15 at T = T0 = 19.89°C mating becomes more efficient and coupling or resting males
as in Table 2, which gives z0v0 = 0.3. The positive activation return to the searching state faster, respectively, so the Allee
energy Ev implies that the encounter rate z increases with threshold declines.
warming. For any pesticide efficiency p, higher temperatures weaken
Other model parameters may respond to warming, too. the Allee effect (i.e. decrease the Allee threshold) via modify-
Experimental results of Thompson et al. (2017) suggest ing the female and male adult emergence distributions and
that the proportion of eggs surviving to the 4th instar or increasing the female−male encounter rate z (Fig. 4b), while
to pupation, which may be positively linked to the param- strengthening the Allee effect (i.e. increasing the Allee thresh-
eter a, might remain unaffected by temperature or decline old) via increasing the background mortality rates dM and dF
with warming. Similarly, the body mass at the 4th instar or (Fig. 4c). The net effect of increasing temperature is then an
the pupal mass, which both may be positively related to the increasingly stronger Allee effect (Fig. 4d). I note that a sin-
number of eggs per egg mass λ, appeared to remain unaf- gle continuous spraying of the infested area by the bacterial
fected by temperature or declined with warming. Moreover, pesticide Bacillus thuringiensis kills about 80% of larvae, i.e.
since activity is expected to increase with warming (Dell et al. p = 0.8 (Blackwood et al. 2012). If the population is at a den-
2014, Osmond et al. 2017), mean resting time tr of just sity, one may eradicate it by increasing the Allee threshold
mated males might increase, too. However, male gypsy moths above that density. Consequently, if for any fixed value of p
mate only during the daylight hours and only once per day the Allee threshold increases with temperature, then at higher
(Blackwood et al. 2012). Therefore, the value of tr = 1 day temperatures the same Allee threshold value can be attained
used in model simulations could stay unaffected by changes with lower pesticide efficiency. Alternatively, even if the
in temperature. population density always remains above the Allee threshold,
For the gypsy moth model, the Allee threshold can be lower perturbations in population density suffice to suppress
defined as the number of egg masses per hectare below the population below it at higher temperatures. A potential
which the population declines toward extinction and decline in the pesticide efficiency p with increasing temper-
above which it grows (Blackwood et al. 2012). In other ature may reverse this pattern. This can be seen in Fig. 4d
words, the population goes extinct if the number of mated when, for example, p = 0.8 at 20°C is reduced to p = 0.6 at
females at the end of mating season is lower than the 30°C. Contrarily, if warming is accompanied by a decrease
number of egg masses at the beginning of the respective in the number of eggs per egg mass and the proportion of
generation and grows if this is not the case. To estimate eggs that survive to become adults in the absence of pesticides
the Allee threshold, I conducted simulations of the gypsy and pupal predation (i.e. by a decrease in the value of λa;
moth model for an increasing number of egg masses and Thompson et al. 2017), the Allee threshold would increase
determined the minimum number of egg masses for which with temperature at a rate higher than in Fig. 4d.
the gypsy moth population grew rather than declined.
Moreover, I determined this minimum number for sev-
eral values of the pesticide efficiency p. This is also because Discussion
pesticide efficiency might depend on temperature, too.
Indeed, the toxicity of two pyrethroids and a spinosyn on Do Allee effects become weaker or stronger when tempera-
the European corn borer Ostrinia nubilalis were found to ture increases? And do Allee thresholds decrease or increase
decrease as post-exposure temperature increased (Musser with warming when the Allee effects are strong? The two
and Shelton 2005). models I examined in this study provide contrasting results.

979
(a) (b) 140
10
10
8 6
4 120
8
100
Protandry: µF − µM

Allee threshold
6 80
95
7 3
60
4
40
2
2 20

0 0
2 4 6 8 10 10 15 20 25 30
Standard deviation: σM = σF Temperature T

(c) 140 (d) 140

120 120
p = 0.8
100 p = 0.6 100
Allee threshold

Allee threshold

p = 0.4 p = 0.8
80 80
p = 0.2 p = 0.6
60 p = 0.0 60 p = 0.4
p = 0.2
40 40 p = 0.0
20 20

0 0
10 15 20 25 30 10 15 20 25 30
Temperature T Temperature T

Figure 4. Dependence of the Allee threshold on (a) phenology of the gypsy moth and (b–d) temperature under various parameters affected
by temperature and several values of the pesticide efficiency p. In panel (b), only the female and male adult emergence distributions (i.e.
protandry µF – µM and maturation dispersion σM = σF; light gray), only the female–male encounter rate z (dark gray), or both adult emer-
gence distributions and z (black) are made temperature-dependent in the gypsy moth model, for p = 0.8 (solid lines), p = 0.6 (dashed lines)
and p = 0.4 (dash-dot lines); I note that p = 0.8 (80% larvae dying due to pesticides) corresponds to a realistic single continuous spraying of
the infested area by the bacterial pesticide Bacillus thuringiensis (Blackwood et al. 2012). In (c), only the male and female background mor-
tality rates dM and dF, respectively, are made temperature-dependent in the gypsy moth model. Panel (d) then demonstrates the net effect of
temperature, since here µF – µM, σM = σF, z, dM and dF are all assumed temperature-dependent. The temperature profiles are calculated from
Eq. 10, 11 and correspond to Logan, WV, USA; the yearly average maximum temperature in Logan is 19.89°C (dotted vertical line in
panels (b–d)). In panel (a), the Allee threshold increases in the direction from dark blue to dark red.

Whereas the Allee effect driven by predation generally weak- mating rate, via increasing female−male encounter rate and
ens and the Allee threshold tends to decrease when tempera- modifying female and male adult emergence distributions.
ture increases, the model with the Allee effect due to need However, it increases male and female mortality rates, too
of finding mates predicts that the Allee threshold should and the net effect is a stronger Allee effect. This summary is
increase with warming. These contrasting results likely reflect overstating the simplicity of the results a bit, since I also show
different mechanisms that trigger individual Allee effects that changes in the predator biomass-density BC in the first
and hence different temperature-dependent parameters. For model and in the pesticide efficiency p in the gypsy moth
the model with predation-driven Allee effect, the metabolic model may under some circumstances reverse these predic-
theory suggests that warming increases the per-unit-bio- tions. Nevertheless, given that there are many aspects of an
mass prey growth rate faster than the predator attack rate. organism’s fitness that might change with temperature, it pro-
Increasing temperature thus weakens the Allee effect. For the vides a neat way of generalizing across these from the two
model with mate-finding Allee effect, warming increases the model examples examined in this study.

980
The assumption of constant predator biomass-density BC population growth rate in the vicinity of Allee threshold
behind the predation model 1 is a reasonable approximation (Walter et al. 2017). It can be thought of as a measure of how
e.g. when the predator’s assimilation efficiency is low and/ fast the population can get away from the Allee threshold
or the predator is a generalist with dynamics largely depen- when it occurs close to it or how resistant it is to repeated
dent on other, unstudied prey species (Gascoigne and Lipcius small-scale perturbations that push the population close to
2004, Berec et al. 2018). Alternatively, one may think of the the Allee threshold (Meyer 2016). The Allee slopes were
predator as a biocontrol agent and of BC as a level of preda- found to increase with warming in the predation model 1,
tor density-biomass maintained by releasing these agents to making smaller Allee thresholds more identifiable and popu-
establish a new or augment a naturally occurring population lations running out of them faster. It is not difficult to show
(van Lenteren 2012). This interpretation makes possible to that for model 1 the Allee slope increases with the square
see both examined models as containing a population control of per-unit-biomass prey growth rate r, which in turn not
means (biocontrol agent versus pesticide) and discuss how its only increases with temperature but has the largest (even in
use affects pest species dynamics when temperature increases. the absolute value) activation energy. Hence, it is r that pri-
In the first model, it is necessary to keep track of decreas- marily determines how the Allee slope responds to warming.
ing Allee thresholds by increasing BC. Similarly, in the second With higher Allee slopes, it may be easier to tune population
model, it is necessary not to let pesticide efficiency deteriorate management around Allee thresholds and better protect the
too much, in view of this possibility being reported (Musser endangered populations as well as better fight the unwanted
and Shelton 2005). This reasoning also indicates that aug- ones. Allee slopes thus appear to be an important characteris-
mentative pest control may perhaps be not as effective as tic of Allee effects (Walter et al. 2017).
using pesticides or disrupting mating, an inference already Up to now, just a single study appears to have examined
made by Blackwood et al. (2012) in a somewhat different an effect of warming on Allee effects. Outcomes of a detailed
context. model describing the life history of invasive cladoceran
It is worth emphasizing that the predictions of the pre- Bythotrephes longimanus suggest that as water temperature
dation model would be reversed (i.e. the Allee effect would will increase the Allee threshold will decrease, facilitating
strengthen with warming) also if the activation energy of the Bythotrephes invasions into non-native habitats as global
predator attack rate a exceeded that of the per-unit-biomass warming proceeds (Wittmann et al. 2011). This prediction
prey growth rate r, Ea > Er . From the pest control (or even falls in line with the results of predation model 1 that Allee
pest eradication) perspective, an increase in the Allee thresh- effects become weaker and the Allee thresholds smaller as
old with temperature may generally be perceived as some- climate change continues. Hopefully further specific as well
thing positive (higher Allee thresholds mean more effective as general models will be developed in which responses of
pest control; Tobin et al. 2011). On the other hand, an Allee effects to warming are under scrutiny. One such ave-
increase in the Allee threshold with warming would imply nue could consider responses of Allee effects to changes in
elevated extinction risk of populations also in their native temperature variability. Indeed, in some environments popu-
habitats where this should be prevented, even if only to pre- lations were found affected more by changes in climate vari-
serve native biodiversity. ability than in climate mean (García-Carreras and Reuman
The findings based on the predation model 1 suggest that 2013) and increases in environmental variance were found
the perception of impacts of climate change depends in part to alter or even reverse effects of changes in environmental
on how one quantifies risk or resilience. On one hand, the mean (Lawson et al. 2015). Despite recent interest in impacts
Allee threshold declines with warming and hence the critical of environmental noise variance and color on the popula-
biomass-density below which the population goes extinct. tion extinction risk (Ovaskainen and Meerson 2010, van
However, at the same time the prey equilibrium biomass- de Pol et al. 2011, Mustin et al. 2013), just a single study
density decreases at a higher rate than the Allee threshold. considered a population subject to an Allee effect. This study
Consequently, their difference, a common measure of popu- showed that as the noise became increasingly autocorrelated
lation resilience (Meyer 2016), declines, too. Perturbations the mean time to population extinction decreased (Levine
from the prey equilibrium biomass-density, likely of increas- and Meerson 2013). However, to make their analysis trac-
ing importance as climate change will continue, may thus table, Levine and Meerson (2013) somewhat unrealistically
threaten a population more as temperature increases despite assumed that the Allee threshold and carrying capacity were
decreasing Allee thresholds. Even though this prediction quite close to one another. Impacts of environmental noise
holds specifically when the prey carrying capacity K declines variance and color on populations subject to an Allee effect
with warming as predicted by the metabolic theory of thus still await exploration.
ecology (Brown et al. 2004), it points to a possibility that
predictions of the effects of climate change on population
dynamics may more generally depend on how we quantify Acknowledgements – Funding – I acknowledge funding by the
risk or resilience. Grant Agency of the Czech Republic (grant no. 15-24456S) and
The Allee slope is a recently introduced Allee effect char- institutional support RVO:60077344 from the Biology Centre.
acteristic that aims to describe behavior of the per capita Conflicts of interest – I declare I have no conflict of interest.

981
References Hutchings, J. A. 2015. Thresholds for impaired species recovery.
– Proc. R. Soc. B 282: 20150654.
Atkinson, D. and Sibly, R. M. 1997. Why are organisms usually Kalinkat, G. et al. 2013. Body masses, functional responses and
bigger in colder environments? Making sense of a life history predator−prey stability. – Ecol. Lett. 16: 1126–1134.
puzzle. – Trends Ecol. Evol. 12: 235–239. Kot, M. 2001. Elements of mathematical ecology. – Cambridge
Bellard, C. et al. 2012. Impacts of climate change on the future of Univ. Press.
biodiversity. – Ecol. Lett. 15: 365–377. Kramer, A. M. and Drake, J. M. 2010. Experimental demonstra-
Berec, L. 2018. Mate search and mate-finding Allee effect: on tion of population extinction due to a predator-driven Allee
modeling mating in sex-structured population models. – Theor. effect. – J. Anim. Ecol. 79: 633–639.
Ecol. 11: 225–244. Lawson, C. R. et al. 2015. Environmental variation and
Berec, L. et al. 2013. Population dynamics of Ips typographus in the population responses to global change. – Ecol. Lett. 18:
Bohemian Forest (Czech Republic): validation of the phenology 724–736.
model PHENIPS and impacts of climate change. – For. Ecol. Levine, E. Y. and Meerson, B. 2013. Impact of colored environmental
Manage. 292: 1–9. noise on the extinction of a long-lived stochastic population:
Berec, L. et al. 2018. Evolution of mate-finding Allee effect in prey. role of the Allee effect. – Phys. Rev. E 87: 032127.
– J. Theor. Biol. 441: 9–18. Meehan, T. D. 2006. Energy use and animal abundance in litter
Binzer, A. et al. 2012. The dynamics of food chains under climate and soil communities. – Ecology 87: 1650–1658.
change and nutrient enrichment. – Phil. Trans. R. Soc. B 367: Meyer, K. 2016. A mathematical review of resilience in ecology.
2935–2944. – Nat. Resour. Model. 29: 339–352.
Binzer, A. et al. 2016. Interactive effects of warming, eutrophication Minder, J. R. et al. 2010. Surface temperature lapse rates over
and size structure: impacts on biodiversity and food−web complex terrain: lessons from the Cascade Mountains. – J.
structure. – Global Change Biol. 22: 220–227. Geophys. Res. 115: D14122.
Blackwood, J. C. et al. 2012. Bioeconomic synergy between tactics Musser, F. R. and Shelton, A. M. 2005. The influence of post-
for insect eradication in the presence of Allee effects. – Proc. R. exposure temperature on the toxicity of insecticides to Ostrinia
Soc. B 279: 2807–2815. nubilalis (Lepidoptera: Crambidae). – Pest Manage. Sci. 61:
Boukal, D. S. and Berec, L. 2002. Single-species models of the Allee 508–510.
effect: extinction boundaries, sex ratios and mate encounters. Mustin, K. et al. 2013. Red noise increases extinction risk during
– J. Theor. Biol. 218: 375–394. rapid climate change. – Divers. Distrib. 19: 815–824.
Brashares, J. S. et al. 2010. Social ‘meltdown’ in the demise of an Osmond, M. M. et al. 2017. Warming-induced changes to body
island endemic: Allee effects and the Vancouver Island marmot. size stabilize consumer–resource dynamics. – Am. Nat. 189.
– J. Anim. Ecol. 79: 965–973. Ovaskainen, O. and Meerson, B. 2010. Stochastic models of
Brown, J. H. et al. 2004. Toward a metabolic theory of ecology. population extinction. – Trends Ecol. Evol. 25: 643–652.
– Ecology 85: 1771–1789. Pereira, H. M. et al. 2010. Scenarios for global biodiversity in the
Courchamp, F. et al. 2008. Allee effects in ecology and conserva- 21st century. – Science 330: 1496–1501.
tion. – Oxford Univ. Press. Robinet, C. et al. 2007. Variation in developmental time
Crates, R. et al. 2017. Undetected Allee effects in Australia’s affects mating success and Allee effects. – Oikos 116:
threatened birds: implications for conservation. – Emu Aust. 1227–1237.
Ornithol. 117: 207–221. Robinet, C. et al. 2008. Dispersion in time and space affect mating
Dell, A. I. et al. 2014. Temperature dependence of trophic success and Allee effects in invading gypsy moth populations.
interactions are driven by asymmetry of species responses and – J. Anim. Ecol. 77: 966–973.
foraging strategy. – J. Anim. Ecol. 83: 70–84. Savage, V. M. et al. 2004. Effects of body size and temperature on
DeLong, J. P. and Hanson, D. T. 2011. Warming alters density population growth. – Am. Nat. 163: 429–441.
dependence, energetic fluxes and population size in a model Sentis, A. et al. 2015. Thermal acclimation modulates the impacts
algae. – Ecol. Complex. 8: 320–325. of temperature and enrichment on trophic interaction strengths
Enquist, B. J. and Niklas, K. J. 2001. Invariant scaling relations and population dynamics. – Global Change Biol. 21:
across tree-dominated communities. – Nature 410: 655–660. 3290–3298.
Fussmann, K. E. et al. 2014. Ecological stability in response to Sentis, A. et al. 2017. Temperature-size responses alter food chain
warming. – Nat. Clim. Change 4: 206–210. persistence across environmental gradients. – Ecol. Lett. 20:
García-Carreras, B. and Reuman, D. C. 2013. Are changes in the 852–862.
mean or variability of climate signals more important for Shaw, A. K. et al. 2018. Sex difference and Allee effects shape the
long-term stochastic growth rate? – PLoS One 8: e63974. dynamics of sex-structured invasions. – J. Anim. Ecol. 87:
Gascoigne, J. C. and Lipcius, R. N. 2004. Allee effects driven by 36–46.
predation. – J. Appl. Ecol. 41: 801–810. Stephens, P. A. and Sutherland, W. J. 1999. Consequences of the
Gascoigne, J. C. et al. 2009. Dangerously few liaisons: a review of Allee effect for behaviour, ecology and conservation. – Trends
mate-finding Allee effects. – Popul. Ecol. 51: 355–372. Ecol. Evol. 14: 401–405.
Gilbert, B. et al. 2014. A bioenergetic framework for the tempera- Stephens, P. A. et al. 1999. What is an Allee effect? – Oikos 87:
ture dependence of trophic interactions. – Ecol. Lett. 17: 185–190.
902–914. Thompson, L. M. et al. 2017. Variation in growth and develop-
Haider, H. S. et al. 2017. Incorporating Allee effects into the mental responses to supraoptimal temperatures near latitudinal
potential biological removal level. – Nat. Resour. Model. 2017: range limits of gypsy moth Lymantria dispar (L.). – Physiol.
e12133. Entomol. 42: 181–190.

982
Tobin, P. C. et al. 2009. The role of Allee effects in gypsy moth, Vasseur, D. A. and McCann, K. S. 2005. A mechanistic approach
Lymantria dispar (L.), invasions. – Popul. Ecol. 51: 373–384. for modeling temperature-dependent consumer−resource
Tobin, P. C. et al. 2011. Exploiting Allee effects for managing dynamics. – Am. Nat. 166: 184–198.
biological invasions. – Ecol. Lett. 14: 615–624. Vercken, E. et al. 2011. Critical patch size generated by Allee effect
Uszko, W. et al. 2017. Effects of warming on predator–prey in gypsy moth, Lymantria dispar (L.) – Ecol. Lett. 14: 179–186.
interactions – a resource-based approach and a theoretical Walter, J. A. et al. 2015. How topography induces reproductive
synthesis. – Ecol. Lett. 20: 513–523. asynchrony and alters gypsy moth invasion dynamics. – J.
van de Pol, M. et al. 2011. Poor environmental tracking can make Anim. Ecol. 84: 188–198.
extinction risk insensitive to the colour of environmental noise. Walter, J. A. et al. 2017. Variation in Allee effects: evidence,
– Proc. R. Soc. B 278: 3713–3722. unknowns and directions forward. – Popul. Ecol. 59: 99–107.
van Lenteren, J. C. 2012. The state of commercial augmentative Wittmann, M. J. et al. 2011. Temperature-dependent Allee effects
biological control: plenty of natural enemies, but a frustrating in a stage-structured model for Bythotrephes establishment.
lack of uptake. – BioControl 57: 1–20. – Biol. Invas. 13: 2477–2497.

Supplementary material (available online as Appendix oik-


05941 at < www.oikosjournal.org/appendix/oik-05941 >).
Appendix 1.

983

You might also like