PROGRESS REPORT – 18 SEPTEMBER 2012
I. Focal Streams
1. Mark Recapture – Vital statistics: The core of our project is four replicate
introduction experiments, each executed in a previously guppy-free headwater tributary
of the Guanapo river. Each replicate was initiated with individually marked guppies
derived from a high predation locality in the lower Guanapo River. We saved three
scales from each individual to provide a source of DNA for genotyping and pedigree
reconstruction. We exhaustively census each locality once per month, identify, measure
and photograph all marked fish, plus photograph, measure and mark all new recruits
(individuals that are >14mm standard length, which is just below the size at which the
fish attain maturity). We initiated the first two introductions in March 2008 and the
second two in March 2009. As of September 5, 2012, we completed our 55th consecutive
monthly sampling. Data has been entered into our on-line data base for the first 54
censuses. We have completed collecting data on length and morphology from the
photographs associated with our first 54 censuses, but have not yet linked the
photographs to our on-line data base. At the end of the 52nd census (summarized in
preparation for the August 2 submission of a NSF grant), we have marked the following
number of individuals:
Lower Lalaja: 6,183 (3,248 females)
Upper Lalaja: 11,188 (5,781 females)
Caigual: 2,874 (1476 females)
Taylor: 5,058 (2,693 females)
Total: 25,303 (13,198 females).
Because we have multiple recaptures for most fish, the accumulated number of
data entries (and associated photographs) were:
Lower Lalaja: 33,577
Upper Lalaja: 60,137
Caigual: 12,476
Taylor: 14,330
Total: 120,520
We add approximately 8,000 new
individuals to the data base per year and 3,500
new data points per monthly census. The figure
on the right illustrates our estimated probably of
recapture if alive. It shows that each census
represents a very high quality sampling of each
population.
What we learn from each census: Each census provides a point estimate of
population size, size structure, per capita recruitment rate, individual growth rate,
population growth rate, and spatial distribution of guppies in the stream. All fish are
collected in a spatially explicit fashion, kept separated by where they were collected
throughout processing, then returned to the point of capture. These data can be combined
with measurements of stream area and topography to provide information on population
density, which can then be integrated with the associated work on the population biology
of the killifish (Rivulus hartii) that is the only other fish species found in this portion of
stream and with the ongoing sampling of the ecosystem.
The underlying mark-recapture statistics reveal information on mortality rate, life
span and, when combined with the pedigree, details on individual reproductive success,
which in turn can lead to a diversity of new questions.
Sample project derived from the monthly censuses: Here we present one example
of the kind of questions that can be addressed with these census data, in addition to
addressing the central goals of the proposal. The following work recently appeared as:
Auer, S. K, A. Lopez-Sepulcre, T. Heatherly, T. J. Kohler, R. D. Bassar, S. A. Thomas
and D. N. Reznick. 2012, J. Anim. Ecol. 81: 818-826. This project synthesized early
results from the mark-recapture study with the periodic ecosystem assessments of
resource availability.
Abstract: 1. Current environmental conditions as well as those in the recent past and
during the juvenile stage can have significant effects on adult performance and
population dynamics, but their relative importance remains unexplored.
2. We examined the influence of food availability in the present, two months prior and
during the juvenile stage on adult somatic growth rates in wild Trinidadian guppies
(Poecilia reticulata).
3. We found that events occurring during early and later parts of an individual’s ontogeny
both had important consequences for adult growth strategies, but the direction of their
influence differed. Low food levels occurring in the present and 2 months prior in the
adult stage both had negative effects on adult growth. In contrast, low food levels during
the juvenile stage generally led to faster adult growth rates. However, these effects of
juvenile food level diminished as the food level experienced two months prior increased.
4. These results suggest that the same conditions at different life stages can have different
effects on the state of the organism and its subsequent short- and long-term growth
strategies. They also suggest that, while juvenile conditions can have lasting effects on
adult performance, the strength of their effects can be dampened by subsequent
environmental heterogeneity in the adult stage.
5. A simultaneous consideration of events in both the adult and juvenile stage past may
therefore improve predictions for individual- and population-level responses to
environmental change.
2. Laboratory life history assays: We continue to assay the life histories of the four
experimental populations once per year. These assays are derived from approximately 40
juveniles collected from each focal stream in February. We begin with small juveniles
because they have only a 10-20% chance of survival to recruitment, so removing them
from the population has much less impact than removing adults. The fish are reared to a
size of 12-15 mm in Trinidad, then transported to our lab in Fort Collins, Colorado
(Cameron Ghalambor, co-PI). It then takes approximately 1 year to rear them to the
second lab-born generation and quantify the life history. These same fish are used to
assess the evolution of morphology and male coloration. We have completed these
assays through the 1/0, 2/1 and 3/2 year cohorts (the first number refers to the number of
years since the first pair of focal streams introductions was initated and the second
number refers to years since the second pair of streams was initiated). We are currently
collecting data from the 4/3 year sample. These assays have shown that males have
evolved later ages and larger body sizes at maturity, as seen in earlier work on this
system and as predicted by life history theory and that all did so within two years of the
introduction.
Project derived from the annual lab life history assays: Here I present one
example of work that has built on the combined information that we are gathering from
comparative studies, our focal stream experiments and the associated laboratory life
history assays. The preliminary work on this project appeared as: Gordon, S. A. LopezSepulcre and D. N. Reznick. 2012. Predation-associated differences in sex-linkage of
wild guppy coloration. Evolution 66: 912-918 (featured by Faculty of 1000)
Abstract: Evolutionary theory predicts that the sex-linkage of sexually selected
traits can influence the direction and rate of evolutionary change, and also itself be
subject to selection. Theory abounds on how sex-specific selection, mate choice, or other
phenomena should favor different types of sex-linked inheritance, yet evidence in nature
remains limited. Here we use hormone assays in Trinidadian guppies to explore the
extent to which linkage of male coloration differs among populations adapted to varying
predation regimes. Results show there is consistently higher degree of X- and autosomal
linkage in body coloration among populations adapted to low-predation environments.
More strikingly, analyses of an introduced population of guppies from a high to a low
predation environment suggest that this difference can change in 50 years or less.
We are currently building on this work by performing an annual assay of the
pattern of color expression in laboratory born females from the focal streams versus the
high predation control site that they were derived from. These assays are giving us a high
resolution record of how the linkage patterns of male colorations changes over time. This
figure shows that there has been an apparent increase in the extent of x-linked cartenoid
coloration within the first year of the introduction.
Carotenoid
HP Canopy Open
Source
Melanistic
HP
Canopy
Source
Open
3. Pedigree update: Our ability to generate an accurate pedigree for the replicate,
evolving populations in the focal streams is essential to multiple goals associated with the
original FIBR proposal. Pedigrees enable us to quantify individual reproductive success,
estimated as the number of offspring produced by each individual, estimated as the
number recruited into the adult population. These data are coupled with our assessment of
individual phenotype, as derived from photographs and to in turn associate reproductive
success with the evolution of key traits, including morphology, size and estimated age at
maturity, and male coloration. The pedigrees also lend themselves to a diversity of other
projects. One example is given below.
State of the Pedigreee: More than 16,000 guppies from the four focal populations
have been genotyped at 10 highly polymorphic microsatellite loci, Pedigree analyses are
conducted using the program Colony [1-2], which uses maximum likelihood methods to
reconstruct nested full- and half-sibship relationships and identify the most likely parents.
Colony’s integrated method of sibship and parentage reconstruction, its sophisticated
approach to dealing with genotyping error (two types of error specified individually for
each locus), and other features, make it the most powerful software available for inferring
parentage in complex systems. We obtained precise genotyping error rate estimates based
on repeated genotyping of hundreds of individuals (>600 replicated observations per
locus). Colony’s analytical power comes at a cost: analysis run times increase greatly
with offspring number, per locus genotyping error, and polygamous mating (exemplified
by guppies), and is also influenced by number of candidate parents and the number of
loci employed. Consequently, Colony runs on monthly batches of guppy recruits can
require up to a week (or more) on a multicore computer. However, pedigree
reconstructions of the first three years (6-9 generations) of one of the focal populations
(lower Lalaja) are illustrative of some of our progress to date. Two parent assignment
rates averaged ~80% over 3345 guppies genotyped, and parentage assignments were
robust, as measured by high levels of concordance obtained from replicate Colony runs
using different random number seeds. However, although average assignment rates were
reasonably high, two-parent assignment rates were negatively correlated with the log of
the number of candidate parents, and as a consequence, declined to 60-70% in the third
year. This decline in assignment rate was not attributable to a loss of genetic variation,
since genetic diversity showed no decline over 3 years (year 1 (3) mean He = 0.766
(0.772); year 1 (3) mean # alleles/locus =12.0 (13.3). Simulations conducted in Cervus
[3] assuming ~3000 candidate parents (typical of the focal populations in their current
state) suggest that our existing panel of 10 loci should be sufficient to confidently assign
parentage to 95% of offspring with greater than 80% confidence. If we increase our
genotyping to 20 loci, then the simulations indicate that we can assign parents to 100% of
offspring with at least 95% confidence. We believe the gap between simulations and our
actual results stem from computational complications imposed by the large number of
candidate parents. Going forward, we plan two remedies: we will make better use of our
resampling data to refine our candidate parental pools, and we will double the number of
loci genotyped. More genetic markers is expected to improve assignment rates, and by
strengthening the genetic ‘signal’, may actually reduce analysis run times (Wang 2012).
Importantly, the fact that no loss of genetic variation has been observed in the focal
populations suggests that ample scope remains for further evolution of these populations.
Project derived from pedigree data – “Beyond lifetime reproductive success. The
posthumous reproductive dynamics of male Trinidadian guppies”. (manuscript
submitted to Proc. Roy. Soc. London, Ser. B on 10 September 2012):
Abstract: Evolution in fluctuating environments requires the long-term
maintenance of genetic diversity in the population, yet selection erodes it. In semelparous
populations, dormant germ banks (e.g. seeds) have been proposed as important in
maintaining genetic diversity and preserving genotypes that are adaptive at different
times. Such hidden storage of genetic diversity need not be exclusive to dormant banks.
Effective population size and genetic diversity may be preserved in many iteroparous
animals through sperm storage mechanisms in females. This allows males to reproduce
posthumously and increase the effective sizes of seemingly female biased populations.
Although long-term sperm storage has been demonstrated in many organisms, the
understanding of its importance in the wild is very poor. We here show the prevalence of
male posthumous reproduction in wild Trinidadian guppies, through the combination of
mark-recapture and pedigree analyses of a multigenerational individual-based dataset. A
significant proportion of the reproductive population consisted of dead males, who could
conceive up to 10 months after death (the maximum allowed by the length of the dataset),
and which is more than twice the estimated generation time. Demographic analysis
shows that the fecundity of dead males has an important contribution to population
growth and selection.
Abundance (# individ. m-2)
4. Impact of guppies on the ecosystem: The
12000
ecosystems ecologists affiliated with the
Total abundance
10000
project have continued with their bimonthly
assessments of the ecosystem in the sections
8000
of the four focal streams where guppies were
6000
introduced and in the control reaches,
upstream of the guppy introduction. Guppies
4000
are excluded from the control reaches by
waterfalls. Thus far, they have documented
Gatherers
8000
that the guppies significantly depress the
LOL control
abundance of invertebrates two years after
LOL introduction
6000
UPL control
their introduction (see Figure on right). This
UPL introduction
figure illustrates the results for just the first
4000
pair of streams that received the guppy
2000
introduction in 2008. “UPL has a thinned
canopy while “LOL” had an intact forest
0
canopy. These results show that the overall
2008
2009
2010
abundance of invertebrates is higher under the
thinned canopy, but that the impact of guppies in invertebrate abundance is similar in
both streams. The second pair of streams generated a similar canopy effect after the first
year. We have yet to complete data collection for 2011 which means that we only have
data for the first year after the guppy introduction in the second pair of streams. As with
the first pair, there was not an impact of guppies on invertebrate abundance after the first
year.
There is a tendency for primary productivity to increase in the guppy introduction
sites relative to the controls, but this result (not illustrated here) falls short of statistical
significance.
Guppies have also significantly depressed the abundance of Rivulus in all four
focal streams. These results are illustrated in the figures on the right. The upper panel
illustrates the population density of Rivulus in
the introduction site minus the Rivulus
population density in control region,
upstream of the guppy introduction site.
Open symbols indicate streams with
experimentally thinned canopies and
closed symbols are the streams with intact
canopies. “LOL” and “UPL” are the pair
of streams for which guppies were
introduced in 2008. Guppies were
introduced to “CAI” and “TAY” in 2009.
This figure illustrates Rivulus population
densities for the 12 month period before
guppies were introduced (-12 to 0 on the xaxis) then for either 48 (LOL and UPL) or
36 (CAI 36
(Cai and TAY) months after the guppies
were introduced. Values less than zero
represent those sampling periods when
Rivulus population densities were higher in
the control region than the introduction
sites. These results show that guppies have
had a
larger impact on Rivulus abundance in streams
with intact
with intact forest canopies. The lower panel
of this figure illustrates the increase in guppy
population density over this same time period.
Guppies depress Rivulus abundance by
constricting the recruitment of juveniles. They do so either by preying on the newborn
Rivulus or competing with the juveniles.
The results for the invertebrates are contained in a paper, senior authored by
Thomas Heatherly (PhD candidate, U. Nebraska), soon to be submitted. The first paper
that reports on the results for Rivulus will be co-authored by Doug Fraser and Brad
Lamphere and is accepted, pending revision for Ecology.
Continued below with “Other ecosystem results” for the focal streams,
No
v
Ja 200
n 7
M 200
ar 8
M 20
ay 08
2
Ju 00
l 8
Se 20
p 08
No 20
v 08
Ja 200
n 8
M 200
ar 9
M 20
ay 09
2
Ju 00
l2 9
Se 0
p 09
20
09
Total invert. abund. (#/m 2)
2
Chlorophyll- (mg/m )
Stage (cm)
100
5. Other ecosystem results – Seasonality
and canopy effects: The two prominent
80
ecosystem effects that emerged from the
study, other than the impact of guppies on
60
the ecosystem, are the effects of the rainy
40
versus dry season and canopy thinning. The
gray bars block out the main dry season,
20
which runs from approximately February
3
until April. This interval represents a time
when flooding events are less likely and in
Pool
2
turn when the abundance of algae and
Riffle
invertebrates increases. The first panel in
1
this figure represents stream depth over time.
The middle panel depicts the standing crop
0
of algae in pools versus riffles. The third
panel represents the total abundance of
-1
invertebrates.
Each large rainfall event scours the
10000
pools and riffles of both algae and
8000
invertebrates. The benthic communities
recover to full abundance over an interval of
6000
approximately 30 days. The difference
4000
between the thinned versus intact canopies
2000
is that primary productivity, and hence the
standing crop of algae and abundance of
0
invertebrates, are higher under the thinned
canopies.
Guppy populations respond to seasonality
by increasing reproductive output during the
dry season and virtually ceasing reprocution in the wet season. The figure on the right
depicts population growth
rate over time from March
2008 through Sept. 2009.
Population growth is
concentrated during the dry
season, when resources are
abundant. Growth is also
more rapid under the
thinned canopy (red line)
again because resources are
more abundant. The figure
below depicts the per capita
recruitment of new
individuals. Peaks that correspond to the main dry season (February through April) and
the lesser dry season in September and October. Recruitment increases during each dry
interval then declines during rainy periods, presumably tracking food availability. This
same pair of figures illustrates the
effects if thinned versus intact
canopies. The red line again
represents the population size and
recruitment in the stream with a
thinned canopy while the black
line represents the data for the
stream that has the intact canopy.
The data are just for the first pair
of streams and represents the time
period from March 2008 through
August 2009. The rate of
recruitment was equal in both sites
for the first dry season, presumably because resources were not limiting in either stream.
The rate of recruitment was higher for the treatment with the thinned canopy in the
remaining two dry seasons, presumably reflecting a significantly higher level of per
capita food availability under the thinned canopy.
Literature Cited
1.
2.
3.
Jones, O.R. and J.L. Wang, COLONY: a program for parentage and sibship
inference from multilocus genotype data. Molecular Ecology Resources, 2010.
10(3): p. 551-555.
Wang, J.L., Computationally efficient sibship and parentage assignment from
multilocus marker data. Genetics, 2012. 166: p. in press.
Kalinowski, S.T., M.L. Taper and T.C. Marshall, Revising how the computer
program CERVUS accommodates genotyping error increases success in paternity
assignment. Molecular Ecology, 2007. 16(5): p. 1099-1106.