---------- Forwarded message ---------From: <proceedingsb@royalsociety.org>
Date: Fri, May 27, 2011 at 7:12 AM
Subject: Proceedings B - Decision on Manuscript ID RSPB-2011-0754
To: tomrevilla@gmail.com
27-May-2011
Dear Dr Revilla:
I am writing to inform you that we have now obtained responses from
referees on manuscript RSPB-2011-0754 entitled "Evolutionary ecology
of seed dispersal by frugivores: The plant's perspective" which you
submitted to Proceedings B.
Unfortunately, on the advice of the Associate Editor and the referees,
your manuscript has been rejected following full peer review.
Competition for space in Proceedings B is currently extremely severe,
as many more manuscripts are submitted to us than we have space to
print. We are therefore only able to publish those that are
exceptional, convincing and present significant advances of broad
interest, and must reject many good manuscripts.
Please find below the comments received from referees concerning your
manuscript, not including confidential reports to the Editor. I hope
you may find these useful should you wish to submit your manuscript
elsewhere.
We are sorry that your manuscript has had an unfavourable outcome, but
would like to thank you for offering your work to Proceedings B.
Sincerely,
Professor Michael Hassell
Editor, Proceedings B
mailto: proceedingsb@royalsociety.org
Associate Editor, Professor Mark Rees
Comments to Author:
We have two detailed reviews of the manuscript, one more theoretically
focused and the other more empirical. Both reviewers feel the analysis
is rather limited and question its relevance to real systems. The
model presented is rather specific and I am not convinced that this
specific scenario is of broad empirical relevance. I am in agreement
with the reviewers and feel we have no option but to reject.
Reviewer(s)' Comments to Author:
Referee: 1
Comments to the Author(s)
I had somewhat mixed impressions of this manuscript. Coming from the
empirical side of research on evolution of “fleshy fruits”, I found it
encouraging with a theoretical analysis of the plant-animal
interactions involved in endozoochory. But my opinion of the ms is
that it is only partly a valuable contribution to this field, and that
it might benefit from reduction in length, and an increase in focus on
the latter part.
The initial part of the paper presents a lengthy exercise in formal
math, based on assumptions that lead to quite trivial conclusions.
Partly, as I understood it, the advantages of dispersal is already
there in the assumptions, implying that there is no real advance in
understanding why these particular dispersal traits are favoured.
Coming as far as through the analysis of density-independent
selection, I had trouble finding anything that was not trivial or
self-evident, even intuitively (skipping the formal math). For
example, selection favours endozoochory if transported seeds have the
highest survival, and the costs of producing fruits are small. If
these costs are instead high, the advantage of endozoochory
disappears. I have also some difficulty of appreciating the value of
conclusions such as “depending on the starting point in the fitness
set, natural selection will lead to the evolution or the elimination
of endozoochory”. A critical question is whether this is a productive
insight. How, for example, could we examine the starting point of the
fitness set in lineages evolving or losing endozoochory? Coming this
far in the ms, I must admit I was quite disappointed.
However, coming over to the section on density-dependent selection,
the analyses become more interesting. This is because the underlying
assumptions are more complex, and the results are not intuitively
obvious. An example is the analysis of the dichotomy where
density-dependence is more, or less, respectively, intense for
transported seeds. In particular, the conditions leading to the
prediction of branching in evolutionary pathways to endozoochory (or
no, or loss of, endozoochory), may stimulate also empirical research.
A challenge may be to examine whether the conditions suggested by the
model conform to “real” cases of origin, or loss of, endozoochory in
different lineages, a topic that has not received much attention (but
see Bolmgren & Eriksson 2005 Oikos 109: 255-272.
In conclusion, this is a timely and motivated theoretical analysis of
evolution of endozoochory. If it is possible, considering the
explanation of the build-up of the model, I suggest a reduction of the
first part (which is mostly trivial). This would probably improve the
value of the paper. In addition, as there are quite many symbols and
abbreviations, it would help to include a table, where all symbols are
explained (it is easy to get lost otherwise).
Minor comment
On Page 3, lines 51 ff, it is stated that generally plants are more
dependent on animals than the other way around. Some papers are cited
in support for this statement. However, I don´t think this is valid
generalization. At least in some tropical systems, a large part of the
fauna is dependent on fleshy fruits (see for example the seminal paper
on the subject: Fleming et al 1987 Ann Rev Ecol Syst). Plants can
always to some extent disperse seeds passively, whereas specialized
frugivores are in trouble without their food source. Without having
any data in support, I would suggest the same holds for birds at
northern latitudes winter-time. For example, waxwings would certainly
die without fruits of rowan (and some other plants). I am not sure
that the populations of the same trees would suffice very much if they
experience a couple of years without any waxwings.
Referee: 2
Comments to the Author(s)
This ms presents and analyzes a model for the evolution of a trait
that effects the mode of dispersal in plants. More specifically, the
evolutionary dynamics of a trait are investigated that affect the
probability that a fruit is eaten by an animal which then disperses
the seeds contained in the fruit.
In my opinion the model analyzes is largely correct, for a few
inconstancies see the comments below.
Remains the question whether the model is interesting. In the
principal, the understanding the forces driving the evolution of
dispersal syndromes is important. What I do not know is to what extent
this question has already been treated from a theoretical perspective
and the ms is of little help here.
The ms makes a few predictions that sound interesting and testable:
e.g., (i) if frugivore seed dispersers are rare, there is selection
for increased investment in conspicuous and rewarding fruits, (ii)
high survival of seeds that are eaten by animals results in increased
selection for frugivory, and (iii) if frugivore animals are easily
attracted, selection favors increased frugivory.
For each of these predictions I could come up with a verbal argument
for why I would expect the opposite pattern, highlighting the
importance of a modeling approach. More specifically, in (i) a
condition that gives frugivory a disadvantage favors increased
frugivory while in (ii) and (iii) conditions that give frugivory an
advantage favor frugivory. This opposing results are a bit puzzling to
me and I do not know what drives them. Thus, in the end I do not have
the feeling that I gained a deeper understanding. The results somehow
come out of the model but I do not learn which aspects of the model
are really responsible from them.
In another prediction of the model, where I actually think I do
understand where it comes from, I do not have much faith. The authors
report that for density-dependent selection evolutionary branching
points exist. At these points increased genetic variation is selected
for resulting in a dimorphism of plants producing attractive fruits
and plants producing inconspicuous fruits. Evolutionary branching
requires negative frequency-dependent selection and as far as I can
see the authors do not discuss where it is coming from. I would like
to offer the following explanation:
The crucial feature of the density-dependent model is that survival of
eaten seeds decreases with the number of eaten seeds while survival of
the non-eaten seeds decreases with the number of non-eaten seeds (see
equation 1). These assumptions imply that eaten seeds are regulated by
a other resources than non-eaten seeds and the authors do not give a
hint what these different regulating factors could be. This needs to
be discussed. I suppose one could argue the following. The
establishment of tree seedlings is primarily limited by open space.
Then, if inconspicuous seeds do not get dispersed at all, they can
essentially only establish if there mother plant dies such that they
can take over its spot. All non-dispersed seeds compete for this spot.
Dispersed seeds can take over empty spots that become available
through the death of all other trees except for the mother plant, not
necessarily of the same species, and all dispersed seeds compete for
this pool of available spots. However, this scenario makes an
assumption where I am not sure it is warranted: seeds that are not
eaten by a frugivore do not get dispersed at all. Frugivory is of
course not the only means of seed dispersal, in particular anemochory
is widespread among trees. Thus, I am not convinced that the most
likely alternative to frugivory is the scenario that seeds drop from
the tree and remain there. And if non-eaten seeds have other means of
dispersal, I do not see where the independent density regulation of
eaten and non-eaten seeds comes from. Eventually, this is an empirical
question. How do the fruits look like from which animal dispersed
fruits were derived?
One other feature of the model that is crucial for the phenomenon of
branching is not discussed. Evolution in the trait z imposes a
trade-off between the fraction of seeds that are dispersed and the
fraction of seeds that are not dispersed. Increasing z, that is, the
attractiveness of seeds to frugivore animals, increases the proportion
of dispersed seeds and decreases the proportion of undispersed seeds.
Negative frequency comes about in the following way. If z is low, most
seeds do not disperse and competition between them is fierce. A mutant
with a higher trait value of z disperses more seeds and these seeds
benefit from weak competition, giving the mutant a rare-type
advantage. An analogous argument applies if z is initially high and we
consider the fate of a mutant with a low value of z. This mutual rare
type advantage allows for branching and coexistence of different types
of seed dispersal strategies. I believe that the fact that increasing
z also imposes a trade-off with the total number of produced seeds, f,
is secondary.
The reason why I have the impression I do not get a deep understanding
of the forces driving seed evolution from the presented model is that
the model is very specific and no effort is made to show the
robustness of the predictions with respect to the various functional
forms. Furthermore, the model is complicated enough such that no
results can be derived analytically. As far as I can see, all results
are derived numerically and since the model already contains a number
of parameters a complete analysis seems almost impossible. In my
opinion, it would be desirable to the see whether it is possible to
derive some results, at least qualitatively, without specifying
explicit functional forms. The aim should be to understand which
requirements are really necessary for a specific pattern. For example,
I have the feeling that the assumption that increasing z decreases f
might be irrelevant for several conclusions of the paper. Similarly,
one might wonder what the impact of density-dependent adult survival
might be of whether increased fruit attractively at the cost of adult
survival would affect the results?
Also, the authors need to make a bigger effort to convince the reader
that they the scenario envisaged by their model is biologically
relevant. How common is it, that trees with seed dispersal by
frugivores derive from trees with basically no dispersal? If this
model is about dispersal vs philopatry, aspects like kin competition
need to be considered explicitly and a huge literature doing this
already exists. If, however, the model should investigate the
evolution of different modes of seed dispersal, this model does not
seem very suitable either.
In section 3.2 the authors analyze a version of their model assuming
density-independent selection. I personally believe that selection is
almost never (if ever) density-independent. So why analyze this
version? The authors do not discuss under what circumstances this
version of their model might be applicable. More sensible would be to
analyze a version where density dependence would be equal for eaten
and non-eaten seeds. Such a model can actually be analyzed by
investigating the optimization criterion given by equation (2) but an
argument has to be made for this (see Metz et al., Evolutionary
Ecology Research, 2008, 10: 629–654).
In the last paragraph of the discussion the authors state that their
results about diversification are in line with results of a paper by
Valido et al with the title "Color, design and reward: phenotypic
integration of fleshy fruit displays". Revilla et al state that that
paper shows how "fundamental differences in the physiology and diet
choices between birds and mammals can explain the differences in
diversification of fruits eaten by these taxa." I have not read that
paper but based on above statement it does not seem to me that the
results of the paper considered here are in line with the paper by
Valido et al since in the present study there is only a single type of
frugivore and diversification is definitely not driven by differences
between different frugivores.
Model Issues
On page 5, line 81, the authors state that the probability that a
fruit escapes consumption is exp(-a*A), where a is the number of
fruits eaten per unit time per frugivore and A is the abundance of
frugivores. With this definition of parameters it seems to me that the
probability to escape consumption should be (number of eaten
fruits)/(total number of fruits)=a*A/(total number of fruits). I do
not see why the exponential distribution comes in. For the chosen
formula to apply it seems necessary that a*A gives probability to be
eaten per unit of time. In conclusion I feel that something is fishy
here.
One reason why the model is so difficult to analyze is the occurrence
of several exponential functions in equation (1). One exponential
could be avoided if the probability to escape frugivory would be
modeled as suggested above. The other exponential comes in through
density dependence where the authors chose a Ricker-type function. I
do not think that there is anything inherent in the life cycle that
makes the Ricker function a more appropriate function than for example
a Beverton-Holt type function for the density dependence.
Avoiding exponential functions in this way do not seem to compromise
the model and has at least two advantages. First, I suspect that then
the viability equilibrium becomes always stable, thus one would not
have to worry about fluctuating population dynamics any longer.
Second, it might even become possible to analyze at least some aspects
of the model analytically since it might become for example possible
to solve for singular points explicitly. However, about this I am not
a hundred percent sure, one would have to try.
The authors give too little information about how results were
derived. For example, how exactly were the PIPs in Figure 4 derived?
Is the equilibrium population size of the resident type calculated
numerically for each x and then checked numerically whether this
equilibrium is indeed stable? I couldn't find information on this.
In line 85 the authors introduce the parameter epsilon, the proportion
of seeds per fruit that survives gut passage. However, a parameter
giving the absolute number of seeds per fruit is missing. Maybe the
model can be scaled such that this number becomes 1 but that should at
least be mentioned.
In equation (2) the authors call the factor that determines population
growth from one time step to the next in the absence of any
competitors R_0. This is a real misnomer. R_0 is widely established as
a symbol for the expected number of offspring over life-time. Since
the authors consider an organism with overlapping generations,
equation (2) clearly does not give R_0 in this sense. An obvious
choice would be to call the expression given on the right-hand side of
equation (2) lambda since it is the dominant eigenvalue of the
one-dimensional system evaluated at the extinction equilibrium.
In line 228-230 the authors describe the fate of a mutant with
positive invasion fitness. This part is sloppy. First, a mutant with
positive invasion fitness (in this model: invasion fitness larger 1)
is not guaranteed to increase in frequency, it might still go extinct
through stochasticity as long as it is rare. Mutants with positive
invasion fitness only have a positive probability to increase in
fitness. Second, a mutant with positive invasion fitness will not
necessarily replace the resident. This is only true for small
mutational steps and residents sufficiently far away from a singular
point. In the neighborhood of a singular point or when a mutant is
sufficiently different from the resident it might actually lie in a
region of coexistence. Similarly, the conclusion in line 257 only
holds if the singular point is also convergence stable.
On page 14 the authors describe the evolutionary dynamics as seen in
individual based simulations. In line 318 they describe how large
mutational steps can alter the predictions derived from adaptive
dynamics. Specifically, in line 320-321 they state that large
mutational steps might lead to a sudden change with one resident type
being replaced by another, very different one. This scenario
corresponds to the PIP shown in Figure 4B which shows alternative
stable states. I disagree with the author that one of the types
present in one alternative stable state is able to displace the other.
In fact, the two types corresponding to the two alternative stable
states are able to coexist, at least on the ecological time scale
(they, however, can collapse into one type on the evolutionary time
scale). This can be seen by mirroring the PIP over the 45-degree line
as described in Geritz et al 1998, and in more detail in Geritz et al.
1999 (TPB). The same applies to the last sentence on this page. Note
that a branching point is not a necessary requirement for two
different phenotypes to be able to coexist on the ecological time
scale. They just have to lie in a "region of mutual invadablility".