Transcript 8, Tupper seminar, larval type and species selection

Causes & consequences of shifts in larval type:
Reconsidering the role of dispersal in
evolutionary processes
Patrick J. Krug
California State University,
Los Angeles
Evolutionary success is unevenly distributed
Major goal of macroevolutionary studies: explain why some
groups are more species-rich than others
3 spp.
Can we identify the traits that explain why
biodiversity is unevenly distributed among
sister clades?
60 spp.
what led this group to out-radiate its sister group
by 20 to 1?
- change in habitat, feeding method, traits involved in
competition or reproduction...?
Evolutionary success is unevenly distributed
Major goal of macroevolutionary studies: explain why some
groups are more species-rich than others
Woo-hoo!
**Winners**
colonizing dry
land led to
explosive
radiations in
many groups
beetles: 350,000 spp.
named (probably >1 million)
Pulmonata: land / freshwater snails + slugs
~60,000 spp. (including marine members)
vertebrates: ~45,000 spp.
Evolutionary success is unevenly distributed
 other lineages can hover at low species numbers despite
being ecologically abundant and important
- may survive unchanged for hundreds of millions of years
and be very well adapted to their niche, yet never diversify
Losers – the “200 club”
sharks + rays: top
marine predators
cephalopods: pinnacle of
invertebrate vision & intelligence
Evolutionary success is unevenly distributed
Diversification rate of a lineage (r) is the net difference between
speciation (new spp. born) and extinction (existing spp. vanish)
- thus, either of two things can lead
to a lineage diversifying more:
1) increase in speciation rate (l)
shift in rate of
diversification
(speciation extinction)
Rabosky 2014
2) decrease in extinction rate (m)
r=l-m
Evolutionary success is unevenly distributed
Diversification rate of a lineage (r) is the net difference between
speciation (new spp. born) and extinction (existing spp. vanish)
- thus, either of two things can lead
to a lineage diversifying more:
1) increase in speciation rate (l)
shift in rate of
diversification
(speciation extinction)
Rabosky 2014
2) decrease in extinction rate (m)
r=l-m
Evolutionary success is unevenly distributed
Diversification rate of a lineage is the net difference between
speciation (new spp. born) and extinction (existing spp. vanish)
1) key innovation may lead to an
adaptive radiation into many
new ecological niches
key innovation
evolves, sets
off burst of
diversification
Rabosky 2014
problem: typically a one-time event,
not naturally replicated
2 living species
of Bosellia
- flat sea slugs
- eat one algal
genus
- tropical only
134 species
in sister clade
Plakobranchidae
- parapodia:
sides rolled up
- eat >20 algal
genera
- tropics to poles
Identifying trait-dependent diversification
Easier to test hypotheses if diversification
rate is character state-dependent,
and character state changes often
ancestral state
derived state
 3x higher rate of
diversification
repeated, independent shifts between
states  naturally replicated experiment
Rabosky
& McCune
2010
Comparative methods can identify such
traits
Identifying trait-dependent diversification
Easier to test hypotheses if diversification
rate is character state-dependent,
and character state changes often
derived state
 3x higher rate of
diversification
Traits that cause greater diversification
result in species selection
- form of selection acting on trait(s)
shared by all members of a species, or
that are a species property (e.g., range)
Rabosky
& McCune
2010
- unrelated to fitness within species
Identifying trait-dependent diversification
l = speciation rate
From a model-fitting perspective,
the question is:
Does a model with two different
speciation rates (one for state
“blue”, one for “red”) fit the data
better than a default model with
the same speciation rate for
both states?
Species selection
in plants
Flowering plants repeatedly
evolved self-compatible
pollen, allowing selffertilization, from selfincompatible pollen
(cannot self-fertilize)
Selfing
Non-selfing
Goldberg et al. 2008, Science
Species selection in plants
In non-selfing plants, estimated speciation rate is higher than
extinction rate – thus, lineages diversify (r > 0)
- however, some non-selfers are always gradually evolving into
self-fertilizers by character change..
selfing
non-selfing
diversification
rate (r)
Goldberg et al. 2008, Science
Species selection in plants
In selfing plants, rates of both speciation and extinction
increase... however, extinction increased more than speciation
- selfing plants have decreased diversification rates (r < 0)
- this explains why non-selfing plants persist, even though some
keep turning into selfers: the remaining non-selfers outcompete
the species that undergo character change and become selfers
selfing
non-selfing
diversification
rate (r)
Marine larval type and dispersal
marine invertebrates produce microscopic larvae that swim for
short periods (0 - 5 days) or long periods (>30 days)
Planktotrophy
lecithotrophy
long-distance
dispersal
short-distance dispersal
Consequences of long-distance dispersal
planktotrophy
lecithotrophy
population
connectivity
gene flow
local adaptation
speciation rate
extinction risk
 planktotrophic populations remain
connected over evolutionary timescales
Evolutionary consequences of larval type
planktotrophy
population
connectivity
gene flow
local adaptation
speciation rate
extinction risk
ancestral
lecithotroph
lecithotrophy
Evolutionary consequences of larval type
planktotrophy
demographic
connectivity
gene flow
local adaptation
speciation rate
extinction risk
populations
diverge...
lecithotrophy
Evolutionary consequences of larval type
planktotrophy
lecithotrophy
demographic
connectivity
gene flow
local adaptation
speciation rate
extinction risk
 theory and genetic data suggest
lecithotrophic populations will split and
diverge into new species...
Evolutionary consequences of larval type
planktotrophy
lecithotrophy
demographic
connectivity
gene flow
local adaptation
speciation rate
extinction risk
 theory and pop-gen data suggest
lecithotrophic populations will split and
divergence into new species...
Evolutionary consequences of larval type
planktotrophy
lecithotrophy
demographic
connectivity
gene flow
local adaptation
speciation rate
extinction risk
 ...but may also go extinct more often
Evolutionary consequences of larval type
For 40 years, paleontological studies of snail fossils have inferred
larval type from the shape of the larval shell, at the tip of adult shell
lecithotrophic shape
Shuto 1974
Shuto 1974, Hansen 1978,
1980, 1982, Jablonski & Lutz
1983, Jablonski 1986
Six studies, cited >1,400 times, concluded lecithotrophs diversify
more than planktotrophs, so benefit from species selection
- that’s 1/12th the number of citations of the discovery of PCR!
Paleontological Perspectives
lecithotrophs
planktotrophs
each vertical line is a species,
showing where it 1st appeared
in the fossil record, and when it
disappeared (went extinct)
 65 million years ago
Hansen 1978, Science
Paleontological Perspectives
lecithotrophic
plankto.
1. lecithotrophs speciate, but also
go extinct, more often
2. planktotrophs survive longer,
speciate less
However, these studies never
calculated diversification rate:
r = speciation - extinction
Hansen 1978, Science
 short-distance may increase
speciation and extinction rates,
but the net difference between
the two is what matters
Paleontological Perspectives
long-distance (n = 50)
%
short-distance (n = 50)
Jablonski (1982, 1986) confirmed
for several groups of snails that
lecithotrophs have higher rates
of both speciation and extinction
 inferred that species selection
favors lecithotrophs, because:
i) they speciate faster
ii) they accumulate in fossil record
over time
%
duration (m. y.)
Has been cited >750 times, and become
a textbook example of species selection
Paleontological Problems
Studies also did not address the fact that short-distance migration
arises in two ways: 1) when a short-distance ancestor speciates, or
2) when a long-distance species undergoes character change
‘species-selection’
hypothesis
short-distance dispersal
evolves once, triggers
rapid diversification
‘character-change’ hypothesis –
accumulation w/out diversification
short-distance evolves 4 times from
different long-distance ancestors;
short-distance species don’t diversify
Molecular phylogenetics & development
Phylogenies of diverse
invertebrate groups show
many independent gains
of lecithotrophy, often with
little subsequent diversification
slipper shells:
- 23 origins of lecithotrophy
- 1 reversal to planktotrophy
- only one radiation (8 spp.)
Collin 2004
Paleontological Problems
i) studies did not factor in rates of character change
ii) paleontological studies never calculated diversification rate
 short-distance may increase both speciation and extinction rates,
but it is the net difference between the two that matters
speciation
rate (l)
diversification rate (r), the rate at
which a lineage accumulates species
(the measure of evolutionary success)
extinction
rate (m)
r = net gain in species over time
Paleontological Problems
i) studies did not factor in rates of character change
ii) paleontological studies never calculated diversification rate
 short-distance may increase both speciation and extinction rates,
but it is the net difference between the two that matters
speciation
rate (l)
planktotrophs
lecithotrophs?
r=1
r=2
extinction
rate (m)
both l and m go up, yet r decreases
Paleontological Problems
long-distance (n = 50)
%
speciation rate (l) = 0.23
extinction rate (m) = 0.17
diversification rate:
(r) = l - m = 0.06
short-distance (n = 50)
speciation rate (l) = 0.43
extinction rate (m) = 0.34
diversification rate:
(r) = l - m = 0.09
%
1) minimal difference (if any...)
duration (m. y.)
Jablonski 1986
2) assumes all “appearances” of shortdistance dispersers reflect speciation,
but some must result from character
change (long turns into short)
Using sea slugs to study macroevolution
Objective: identify traits that promote diversification, using
herbivorous slugs in clade Sacoglossa as a model
Using sea slugs to study macroevolution
Several facets of sacoglossan biology suggest development mode
is evolutionarily free to evolve in this group:
- includes 5 of 8 species in which development
mode varies within a species, due to
dimorphism in egg size
- sister species often differ in development
- per-offspring investment varies widely among
species, with some
investing heavily in
extra-capsular yolk
(similar to nurse eggs)
4 gene phylogeny of Sacoglossa
- data: mtDNA: COI, 16S (1,062 bp)
nuclear: H3, 28S (1,720 bp)
- Maximum Likelihood (RaxML)
- one data partition, GTR + Γ
- Bayesian Inference with 3 mixture
models (BayesPhylogenies), 108
generations, 4 independent runs
- all nodes shown had >90% support (BI)
- 202 ingroup species (74 undescribed)
- developmental data for 114 spp.
Oxynoacea - 6 genera, 74 spp.
shelled
Limapontiodea - 18 genera, 152 spp.
cerata-bearing
Plakobranchoidea
- 4 genera, 137 spp.
(103 in Elysia)
photosynthetic
short-distance
long-distance
Ancestral devel. mode
Limapontioideainferred using
evolutionary quantitative
probability
genetics
that an
model
ancestor had
given type of larval dispersal
a
short-distance
long-distance
Plakobranchoidea
- more lecithotrophs in Plakobranchoidea, but
only two pairs of lecithotrophic sister species
 species-selection hypothesis predicts (a) clades of short-distance
dispersers, which (b) should contain more species
 NOT the case!
1. Testing for shifts in
diversification rate
Software ‘Medusa’ used to model
diversification across 32 genus-level
clades, using total # of known spp.
1
Medusa identifies shifts in the overall
rate of diversification, not taking into
considerating character state
2
 two branches where rate of
diversification accelerated:
1) after loss of shell
2) after photosynthesis evolved
Alfaro et al. 2008
2. Testing state-dependent
diversification rate
BiSSE used to model rates of speciation,
extinction, and character change
Tested fit of alternative models with:
a) 2 rates, depending on larval type
or
b) one rate, independent of larval type
Considered the three superfamilies of
Sacoglossa as distinct, since they
diversify at different background rates
Maddison et al. 2007, FitzJohn 2012
Speciation rate depends on larval type
a) unrestricted BiSSE
l (1), m (1), q (1)
l (1), m (2), q (1)
l (2), m (1), q (1)
l (2), m (2), q (1)
df
ln(L)
AIC
χ2
P
9
12
12
15
-68.73
-66.06
-61.90
-61.20
155.46
156.12
147.79
152.40
n/a
5.33
13.67
15.06
n/a
0.149
0.003
0.020
best-fit
model
b) restricted BiSSE
l- (1),
m (1),which
q (1) allowed
9
-70.33
158.66
n/a vary n/a
model
speciation
rate to
with larval type
l (1),
m (2),
q (1) preferred
12
-66.87
157.75 which
6.91 ignored
0.075 larval type
was
highly
over model
l (2), m (1), q (1)
12
-63.78
151.55
13.10
0.004
l (2), m (2), q (1)
15
-63.29
156.58
14.08
0.029
- letting extinction rate covary with larval type did not improve fit
l = speciation rate
m = extinction rate
q = rate of character change
Krug et al., Systematic Biology, in press
Speciation rate depends on larval type
df
ln(L)
AIC
χ2
P
a) unrestricted BiSSE
l (1), m (1), q (1)
l (1), m (2), q (1)
l (2), m (1), q (1)
l (2), m (2), q (1)
9
12
12
15
-68.73
-66.06
-61.90
-61.20
155.46
156.12
147.79
152.40
n/a
5.33
13.67
15.06
n/a
0.149
0.003
0.020
b) restricted BiSSE
l (1), m (1), q (1)
l (1), m (2), q (1)
l (2), m (1), q (1)
l (2), m (2), q (1)
9
12
12
15
-70.33
-66.87
-63.78
-63.29
158.66
157.75
151.55
156.58
n/a
6.91
13.10
14.08
n/a
0.075
0.004
0.029
best-fit
model
- same result whether reversals to planktotrophy were free to
occur (top), or were constrained to be very rare (bottom)
But which larval type actually
benefited from species selection??
Species selection favors planktotrophy
plankto
rP
Oxynoacea
Limapontioidea
Plakobranchoidea
3.2
10.4
26.1
lecitho
rL
1.8
<0
10.1
q1
4.3
1.1
9.8
 diversification rate (speciation – extinction) was always higher
for planktotrophs (rP) than lecithotrophs (rL)
1. Oxynoacea: overall diversification was low, but rP was ~twice rL
Species selection favors planktotrophy
plankto
rP
Oxynoacea
Limapontioidea
Plakobranchoidea
3.2
10.4
26.1
lecitho
rL
1.8
<0
10.1
q1
4.3
1.1
9.8
 diversification rate (speciation – extinction) was always higher
for planktotrophs (rP) than lecithotrophs (rL)
2. Limapontioidea: only planktotrophs diversified! (rL<0)
- lecithotrophs arose exclusively from character change
Species selection favors planktotrophy
plankto
rP
Oxynoacea
Limapontioidea
Plakobranchoidea
3.2
10.4
26.1
lecitho
rL
1.8
<0
10.1
q1
4.3
1.1
9.8
 diversification rate (speciation – extinction) was always higher
for planktotrophs (rP) than lecithotrophs (rL)
3. Plakobranchoidea: high rates of overall diversification for
photosynthetic clade, and frequent character change
 however, rP was still ~twice rL
Are sacoglossans just weird, though?
“This is not a group that appears to have speciation rates driven
by lecithotrophy: lecithotrophy is the much rarer state in this
group. Presumably this is not the case for many other clades.”
“You are characterizing patterns in a single somewhat odd
clade of mollusks, with relatively poor fossilization.”
- anonymous reviewer comments about this work
Are sacoglossans just weird, though?
Heterobranchia
Anaspidea
Cephalaspidea
Notaspidea
Nudibranchia
Sacoglossa
#P
17
47
7
171
108
#L
2
13
3
60
35
%P
89.5
78.3
70.0
74.0
75.5
Caenogastropoda
Calyptraeidae
Conidae
Fasciolariidae
Littorininae
Muricidae
Volutidae
39
56
9
139
36
0
39
35
25
13
46
9
50.0
61.5
26.5
91.4
43.9
0.0
% of known species
with planktotrophic
development
outliers are some clades
in Neogastropoda that
have few surviving
planktotrophs
...but guess who paleontological studies focused on?
Are sacoglossans just weird, though?
“This is not a group that appears to have speciation rates driven
by lecithotrophy: lecithotrophy is the much rarer state in this
group. Presumably this is not the case for many other clades.”
“You are characterizing patterns in a single somewhat odd
clade of mollusks, with relatively poor fossilization.”
As a function of changes per branch, larval type changed
about as often in Sacoglossa (0.067) as in cone snails (0.067),
and less often than in slipper shells (0.176)
Thus, Sacoglossa is typical in its % of planktotrophs, and in
its rate of developmental evolution
Tempo and mode of larval evolution
Changes in a trait like larval type can be:
a) strongly affected by phylogenetic
relationships – close relatives tend
to share similar traits
- a pre-requisite for species selection
to favor lecithotrophy is that it must
be heritable at the species level
- i.e., lecithotrophs must give rise to
daughter species that are also
lecithotrophic
Tempo and mode of larval evolution
Changes in a trait like larval type can be:
a) strongly affected by phylogenetic
relationships – close relatives tend
to share similar traits
test: compare model fit allowing egg size
to evolve across tree with vs. without
Pagel’s scaling parameter l
- reflects the expected covariance of
close relatives, who should have similar
egg sizes
l= 0, relatives aren’t similar in egg size
l = 1, relatives share egg size (larval type)
Tempo and mode of larval evolution
Changes in a trait like larval type can be:
a) strongly affected by phylogenetic
relationships – close relatives tend
to share similar traits
result:
l= 0
l= ML value
(0.38 ± 0.18)
model
likelihood
-428.14
BF
test
-428.01
0.25
 BF <2 means, no support for model
that corrects for relatedness
 relatives are not similar in larval type
Tempo and mode of larval evolution
Changes in a trait like larval type can:
b) occur early or late in history of a group
- early change = adaptive radiation,
followed by slow-down in evolution
- late change = species-specific adaption
occurring near tips of the tree
(change is disproportionately recent)
Tempo and mode of larval evolution
Changes in a trait like larval type can:
b) occur early or late in history of a group
test: compare model fit allowing egg size
to evolve across tree with vs. without
Pagel’s scaling parameter d
d < 1, most change occurs on short
branches (near root of tree)
d = 1 (default), change in larval type
equally likely anywhere on tree
d > 1, change in larval type gets more
likely as branches get longer
(most change is near tips)
Tempo and mode of larval evolution
Changes in a trait like larval type can:
b) occur early or late in history of a group
result:
d=1
d = ML value
(2.8 ± 0.2)
model
likelihood
-448.42
BF
test
-435.46
25.9
 BF >5 = strong support for model
with change concentrated near tips
 larval type changes as species-specific
adaptation, recently in history
Short-term solutions to a long-term problem
Species selection favors long-distance dispersal in Sacoglossa,
and perhaps in most invertebrate groups
Loss of dispersive larvae is..
i) favored at ecological timescales, so change is frequent
ii) a dead-end at macro-evolutionary timescales
Most short-distance
dispersers are
the Walking Dead:
short-lived lineages that
go extinct before they
can diversify into
daughter species
Then why does lecithotrophy evolve so often?
A trait that confers high fitness within species (individual-level)
doesn’t necessarily produce evolutionary success (speciation)
1) studies of poecilogonous species (variable development) indicate
planktotrophy is selected against when dispersal is unlikely to
succeed due to oceanographic constraints
- Alderia willowi expresses lecithotrophy when Californian
estuaries close during the dry season (Krug et al. 2012)
- C. ocellifera is lecithotrophic in enclosed Caribbean lagoons
(Ellingson & Krug, in revision)
 selection may be imposed by larval environments that make
planktonic dispersal unlikely to succeed
Then why does lecithotrophy evolve so often?
A trait that confers high fitness within species (individual-level)
doesn’t necessarily produce evolutionary success (speciation)
2) Models of correlated trait evolution revealed that increased
per-offspring investment in planktotrophs increased rate at
which lecithotrophy evolved in Sacoglossa (Krug et al., in press)
 adult reproductive traits may impose correlated selection on
larval type, which affects fecundity
Short-term solutions to a long-term problem
Species selection favors:
1) self-incompatible pollen in plants
2) long-distance larval dispersal in molluscs
 in both cases, the derived state (selfing in plants, short-distance
larvae in sea slugs) evolves frequently, but increases
extinction more than speciation, so dooms that lineage
 thus, what’s favored by selection in the short-term or within a
species may not be an evolutionarily “winning strategy” in the
long term
>1,400 citations support a hypothesis that our results indicate is
wrong. Don’t believe everything you read!
Thanks...
Jann
Vendetti
DEB (Systematics)
OCE
Ryan
Ellingson
Diane
Rico
Cynthia Trowbridge
Oregon Inst. Marine Biology
Julia Vo
Angel Valdes
Cal Poly Pomona
Nerida Wilson
Australian Museum
Danielle
Trathen
Elena
Hidalgo
Albert
Rodriguez