**Female and male time budgets **

We assume that the number of males and their mate searching ability means that every female is guarded at the time of the siring opportunity. Below we show why this

assumption remains valid at the evolutionary equilibrium even if the ASR is female-biased.
Females spend a fraction *b* of their time being precopulatorily guarded, a fraction *a* of their
time being postcopulatorily guarded, and the remaining time, 1–*a*–*b*, they spend

unguarded (Figure 2). The number of unguarded females (*f*) is therefore, on average,

*f *= *F* (1–*a*–*b*) (S3)

While females can breed every 1 unit of time, this is not true for males because *M* does not
necessarily equal *F*. A female who is *b* time units away from breeding (i.e., has been

unguarded for 1–*a*–*b* units of time) is guaranteed to become guarded by a male. Males, if
they are numerous (*M* > *F*), have to wait on average longer than 1–*a*–*b *to find a female
who is *b* time units away from giving the siring opportunity. Conversely, if there are
relatively few males (*M* < *F*), their wait time is shorter than 1–*a*–*b*. We denote the time a
male spends on roaming by *T*, thus the length of a breeding cycle for males is, on average,
*T*+*a*+*b*. The number of non-guarding males (*m*) in the population is given by

𝑚 = 𝑇

𝑇+𝑎+𝑏𝑀 (S4)

As we did in the non-cue model (Supplement I), we specify *p*, the number of pairs

(guarding male + guarded female) in the population: this obeys *m* = *M* – *p* and *f* = *F* – *p*, and
can also be expressed as *p* = *F* (*a*+*b*). From (S4) we therefore find the solution for the
average male wait time,

𝑇 =𝑀_{𝐹}− 𝑎 − 𝑏 (S5)

Note that our model is applicable as long as *T* ≥ 0 (i.e., the evolution of *a* and *b* cannot
proceed beyond the constraint *a*+*b* < *M*/*F*); as we will see, this constraint will be
automatically satisfied by evolutionarily stable behaviour.

To derive average male fitness in a population where guard durations *a* and *b* are in use,
we note that a male breeding cycle consists of guarding (duration *a*+*b*), during which the
male gains fitness

and a period of non-guarding (duration *T*), during which he can use the siring

opportunities that postcopulatorily unguarded females offer. These benefits to ‘roaming’
males are of magnitude 1 –* x* = *S*I + *Se*–*ka*, and they are offered by *f* females, distributed over
a time of duration 1–*a*–*b*, and competed over by *m* males. Thus, the fitness accrued during
a roaming period of length *T* is *Ty*, where

𝑦 = 𝑓

𝑚

𝑆𝐼+𝑆𝑒−𝑘𝑎

1−𝑎−𝑏 (S7)

is the fitness gain per unit time when roaming, i.e., fertilizing postcopulatory unguarded
females’ eggs. Using known relationships between *F*, *M*, *f* and *m* (above), this can also be
expressed as

𝑦 =𝑆𝐼+𝑆𝑒−𝑘𝑎 𝑀

𝐹−𝑎−𝑏

(S8)

Male fitness per unit time of his life is

𝑊 = 𝑥+𝑇𝑦

𝑇+𝑎+𝑏 (S9)

Using equations (S3)-(S6) and (S8), this expression simplifies to

𝑊 = 𝐹

𝑀 (S10)

This makes good sense. One male or another gets a total of 1 unit of siring opportunities
for each breeding cycle of each female, thus for the average male, the total number of
siring opportunities that arise for him, either while guarding or non-guarding, must equal
*F*/*M*.

**Fitness of a mutant male **

The above result refers to the population average when all males use *a* and *b* as pre- and
postcopulatory guarding durations. We now ask what happens when a male deviates from
the norm and uses a different value of *a* and/or *b* from the rest of the population.

We deal with postcopulatory guarding first, as this is the simpler case. Altering this
duration, *a*, impacts the rate at which a mutant male completes his breeding cycle without
altering his wait time (which remains the population average, equation S5). Changing the
guarding duration also changes the proportion of eggs fertilized that are made available by
the guarded female. Thus, a male with a deviant value of *a* has his fitness characterized not
by eq. (S10) but by (S9) with *T *and *y* following the population norm, and *a* and *x* having a
different value from the resident strategy. Differentiating, we obtain

𝜕𝑊
𝜕𝑎 =
𝜕
𝜕𝑎
𝑆𝐵+𝑆(1−𝑒−𝑘𝑎)+𝑇𝑦
𝑇+𝑎+𝑏 =
𝐹
𝑀(𝑆𝑘𝑒
−𝑘𝑎_{−}𝐹
𝑀) (A11)

This is positive when 𝑆𝑘𝑒−𝑘𝑎_{>} 𝐹

𝑀, which means that *a* will increase until

𝑎∗_{=}ln(𝑆𝑘
𝑀

𝐹)

𝑘 (S12)

Note that if *SkM*/*F* < 1, 𝜕𝑊

𝜕𝑎 is negative for all *a*, thus *a** = 0. Thus, a small scope for
paternity improvement, a long fertilization window (small *k*), and a female-biased sex
ratio all decrease the likelihood that any postcopulatory guarding will occur.

Fitness calculations for precopulatory guarding are slightly different than those for
postcopulatory guarding, because a shorter or longer *b* cannot be assumed to lead to a
different duration of guarding or roaming until a new mate is found, respectively. Instead,
given that we assume sufficiently many males that each female will be guarded every
breeding cycle, a male who delays guarding (uses a shorter *b*) will never find a female
approaching her *t* = 0. Thus a male with a shorter *b* than the population average roams
continually, gaining fitness through inefficiently guarded eggs as well as those still
unfertilized after another male’s guarding period is over. A roaming male’s fitness
accumulates at a rate *y* per time unit. It follows that if *y* > *F*/*M* (where *y* follows equation
S8), shorter values of *b* are favoured.

An increased value of *b* has similarly drastic effects on mate-finding. A male whose *b*
exceeds the population average by an infinitesimally small duration ∆*b* can remove all his
waiting time: he accepts females further away from giving a siring opportunity than what
is acceptable to other males, thus in the absence of competition, he is guaranteed to find
such a female without delay. A male who avoids all waiting by shifting to *b*+∆*b* (where ∆*b*
is approximately zero) has fitness 𝑆𝐵+(𝑆)(1−𝑒−𝑘𝑎)

𝑎+𝑏 obtained from (S9) with waiting time set
to zero. Thus, longer *b* values are favoured when 𝑆𝐵+(𝑆)(1−𝑒−𝑘𝑎)

𝑎+𝑏 > 𝐹

𝑀, and shorter *b* values
are favoured when *b* leads to *y* > *F*/*M*. It follows that selection favours longer guarding *b*
when *b* is ‘too small’, shorter values of *b *when *b *is ‘too long’, and these two options collide
when
𝑆𝐵+(𝑆)(1−𝑒−𝑘𝑎)
𝑎+𝑏 =
𝐹
𝑀= 𝑦 =
𝑆𝐼+(𝑆)𝑒−𝑘𝑎
𝑀
𝐹−𝑎−𝑏
(S13)

Biologically, the interpretation is that when *b* increases in the population, competitor
males spend longer in each breeding cycle, which means that fewer males compete for the
eggs that guarding males cannot fertilize (those that have been left unguarded and are still
not fertilized, or those that remain unprotected by inefficient guarding). This frequency

the point where non-guarding leads to the same fitness as guarding for an infinitesimally longer period than the population average, we have the equilibrium.

𝑏∗_{= {}
𝑀
𝐹(1 − 𝑆𝐼− 𝑆𝑒
−𝑘𝑎_{) − 𝑎𝑖𝑓𝑎}∗_{> 0}
𝑀
𝐹𝑆𝐵𝑖𝑓𝑎
∗ _{= 0} (S14)

The sum *a**+*b** (eq. A12 + eq. A14), in cases where guarding occurs, is 𝑀

𝐹(1 − 𝑆𝐼) −
1
𝑘,
which gives an intuitive first result: the duration of guarding as a whole increases with the
ASR and decreases if guarding is inefficient (high *S*I) ; it also increases with the speed of
eggs becoming fertilized. We are also finally able to confirm our assumption, that there
will always be enough males to begin guarding females who are *b** time units away from
offering a siring opportunity, is valid. This follows because the waiting time, 𝑇 =𝑀_{𝐹}− 𝑎 −
𝑏, reaches the equilibrium value 𝑀𝑆_{𝐹} +1_{𝑘} if *a** > 0, and 𝑇 =𝑀_{𝐹}(1 − 𝑆𝐵) if 𝑎∗= 0. Since both
values are positive, the roaming population of males does not ‘run out’ under either
scenario: each male still spends some time *T* > 0 roaming in between two guarding bouts.
There are two special cases. We already noted above that postcopulatory mate-guarding is
not selected for when𝑆𝑘𝑀

𝐹. Second, it is also possible that the sum *a**+*b** ≥ 1. This means
that males are never selected to stop guarding a single focal female; they persist in
guarding from one breeding cycle to the next. This solution implies male monogamy.

**Chapter 2 – Demography can favour female-advantageous **

**alleles **

Anna MF Harts, Lisa E Schwanz and Hanna Kokko

* Proceedings of the Royal Society of London B*(2014) 281: 20140005

**Abstract **

When female fecundity is relatively independent of male abundance, while male reproduction is proportional to female abundance, females have a larger effect on population dynamics than males (i.e. female demographic dominance). This population dynamic phenomenon might not appear to influence evolution, because male and female genomes still contribute equally much to the next generation. However, here we examine two evolutionary scenarios to provide a proof of principle that spatial structure can make female demographic dominance matter. Our two simulation models combine dispersal evolution with local adaptation subjected to intralocus sexual conflict and

environmentally driven sex ratio biases, respectively. Both models have equilibria where one environment (without being intrinsically poorer) has so few reproductive females that trait evolution becomes disproportionately determined by those environments where females survive better (intralocus sexual conflict model), or where daughters are

overproduced (environmental sex determination model). Surprisingly, however, the two facts that selection favours alleles that benefit females, and population growth is improved when female fitness is high, together do not imply that all measures of population

performance are improved. The sex-specificity of the source–sink dynamics predicts that populations can evolve to fail to persist in habitats where alleles do poorly when

expressed in females.