Animal Biodiversity and Conservation 42.

2 (2019) 293

Survival and longevity in neotropical

damselflies (Odonata, Polythoridae)

A. Cordero–Rivera, I. Sanmartín–Villar,
M. Sánchez Herrera, A. Rivas–Torres,
A . C. Encalada

Cordero–Rivera, A., Sanmartín–Villar, I., Sánchez Herrera, M., Rivas–Torres, A., Encalada, A. C., 2019. Sur-
vival and longevity in neotropical damselflies (Odonata, Polythoridae). Animal Biodiversity and Conservation,
42.2: 293–300, Doi: https://doi.org/10.32800/abc.2019.42.0293

Abstract
Survival and longevity in neotropical damselflies (Odonata, Polythoridae). Longevity among insect orders va-
ries greatly, and has mainly been studied in insects in temperate biomes, where seasonality determines high
synchronization of reproductive activities and limits lifespan. Most forest damselflies in tropical regions have
low population densities and are almost never observed in copula. We hypothesized that selection will favour
a high survival rate and hence high lifespan, allowing the animals to be ready for the occasional events that
favour reproduction. We studied two neotropical damselflies, Polythore mutata and P. derivata, in Ecuador,
using mark–recapture methods. We found that sex affected the rate of recapture, but daily survival rate was
affected by sex only in one population. We found evidence that suggests stabilizing or directional selection on
body size. The maximum lifespan was 54–63 days. We conclude that the survival rate of Polythore damselflies
in tropical forests is comparable to that of similar damselflies in temperate zones.

Key words: Lifespan, Rainforest, Low density, Body size, Mark–recapture

Resumen
Supervivencia y longevidad de las libélulas del neotrópico (Odonata, Polythoridae). La longevidad, que entre los
órdenes de insectos es muy variable, se ha estudiado principalmente en insectos de biomas templados, donde
la estacionalidad determina una alta sincronización de las actividades reproductivas y limita la longevidad. La
mayoría de las libélulas de las regiones tropicales vive en poblaciones con una densidad baja y casi nunca
se observan en cópula. Nuestra hipótesis es que la selección favorecerá una alta tasa de supervivencia y,
por lo tanto, una gran esperanza de vida, lo que permitiría que los animales estuvieran listos para los even-
tos ocasionales que favorecen la reproducción. Estudiamos dos libélulas neotropicales, Polythore mutata y
P. derivata, en Ecuador, utilizando métodos de marcaje y recaptura. Constatamos que el sexo afectó a la tasa
de recaptura, pero que solo afectó a la tasa de supervivencia diaria en una población. Hallamos indicios que
sugieren la existencia de selección estabilizadora o direccional del tamaño del cuerpo. La longevidad máxima
observada fue de 54–63 días. Concluimos que la tasa de supervivencia de las libélulas del género Polythore
en los bosques tropicales es comparable a la de libélulas similares de las zonas templadas.

Palabras clave: Longevidad, Selva tropical, Baja densidad, Tamaño corporal, Marcaje y recaptura

Received: 30 X 18; Conditional acceptance: 11 I 19; Final acceptance: 12 III 19

Adolfo Cordero–Rivera, Iago Sanmartín–Villar, Anais Rivas–Torres, ECOEVO Lab, Escola de Enxeñaría
Forestal, Universidade de Vigo, Campus A Xunqueira, 36005 Pontevedra, Galiza, Spain.– Melissa Sánchez
Herrera, Biology Program, Faculty of Natural Sciences and Mathematics, Universidad del Rosario, Bogotá,
Colombia.– Andrea C. Encalada, Laboratorio de Ecología Acuática, Instituto BIOSFERA, Universidad San
Francisco de Quito, Diego de Robles y Vía Interoceánica, Campus Cumbayá, 17–12–841, Quito, Ecuador.

Corresponding author: A. Cordero–Rivera. E–mail: adolfo.cordero@uvigo.es

ISSN: 1578–665 X © 2019 Museu de Ciències Naturals de Barcelona

eISSN: 2014–928 X Papers are published under a
Creative Commons Attribution 4.0 International License
294 Cordero–Rivera et al.

Introduction in Ecuador. This author suggested that P. mutata

phenotypes converge in two of the seven mimicry
Over the last decades, ecological research on natu- rings he described for this geographic location. His
ral populations of insects has accumulated a large work was mainly descriptive, but he also looked at
dataset about survival and reproductive behaviour, the UV reflections which seemed to be consistent
allowing a good understanding of their reproductive with the butterflies as well. Other researchers have
strategies (Cornell and Hawkins, 1995). Longevity and anecdotally reported the similarities of the flight of
fecundity among insect orders vary greatly, and recent Polythore damselflies with butterflies, and recent stu-
evidence indicates that some insects live long enough dies of the flight behaviour in the species Polythore
to experience reproductive senescence (Sherratt et procera, Euthore fasciata and Gretta andromica
al., 2010). However while strong statistical evidence suggest that there is a mimicry signal among them
has been presented for senescence in vertebrate (Outomuro et al., 2016a) wing shape, and flight
populations in the wild, we know little about the rate style. The study species have wings with a subapical
and shape of senescence in wild populations of in- white patch, considered the aposematic signal, and
sects. Most of this information comes from studies of a more apical black patch. The main predators are
insects in temperate biomes (Cordero–Rivera and VS–birds, visually more sensitive to violet than to
Stoks, 2008), where seasonal regulation (Tauber ultraviolet wavelengths.
and Tauber, 1976) and reproductive synchronization Using multiple mark and recapture methods, we
(Cordero Rivera and Andrés Abad, 1999) are clearly estimated recapture rates, survival probability and life
adaptive. Nevertheless, in tropical regions, where expectancy of two geographically distant populations
temperature remains almost constant, there are no of P. mutata in Ecuador. Our hypothesis was that these
cues to reproductive synchronization, except those animals have high survival (see above). We aimed to
derived from rain regimes (Wikelski et al., 2000). Odo- answer the following questions: 1) is the survival rate
nates (damselflies and dragonflies) have been widely different between males and females?; 2) are both
studied using mark–recapture methods in temperate sexes recaptured with similar probabilities?; 3) for P.
regions (Cordero-Rivera and Stoks, 2008). For these mutata females, is the survival probability of colour
reasons, studies of longevity of tropical damselflies morphs the same?; and 4) does body length affect
are clearly needed (Cordero–Rivera and Stoks, 2008). survival? Given the presence of a small population
In tropical forests, some dragonflies are key species of Polythore derivata in one of the field sites, we also
linking stream trophic networks to aerial networks, give a first analysis of survival in this species.
and are present at very low densities, particularly
in South America. For instance, individuals of the
family Calopterygidae can be found in their thou- Material and methods
sands in European streams but only a few scattered
specimens can be found in the Amazon (for instance Field data collection
members of the genera Hetaerina, Mnesarete or
Ormenophlebia) (Córdoba–Aguilar and Cordero–Rive- The field data collection and observations were con-
ra, 2005). In these tropical areas, reproductive activity ducted at two localities in Ecuador where P. mutata
is rarely observed. We hypothesized that animals must was previously recorded. The first location was the
survive long periods to be able to reproduce when Tiputini Biological Station (TBS), at the border of Yasu-
conditions are appropriate (e.g. Sanmartín–Villar and ni National Park (76.146041ºW, 0.635000ºS, Orellana
Cordero–Rivera, 2016). province). We visited this location three times; from
Due to its particular individual and population traits, 4 to 12 December 2012, from 4 to 7 February 2013,
the family Polythoridae (Zygoptera) is a relevant and from 10 to 23 June 2013. However, at the third
model in ecology and evolution (Sánchez Herrera date, the density of P. mutata was extremely low and
et al., 2015). A recent molecular phylogeny of this no marked individuals were resighted, which suggests
family suggests the existence of two clades, one that the time span of the study was appropriate for
grouping the species distributed across the Amazon the expected lifespan of the species. A few individuals
basin and the other grouping species in the Andes of Polythore derivata were observed during these
(Sanchez Herrera et al., 2018). Most species of this sampling periods, but too few to be studied.
genus possess sexual dimorphism, with males and The second field site was the Jatun Sacha Bio-
females displaying different colour patterns. For logical Station (JSBS), near Tena (77.615677 ºW,
example, the Amazonian species Polythore mutata 1.067593 ºS, Napo province), the same place visited
(McLachlan, 1881) shows female colour polymor- by Beccaloni (1997). Here the density of P. mutata
phism in addition to sexual dimorphism. The males was higher than in Tiputini, and was accompanied by
and one phenotype of females (androchrome, similar some Polythore derivata and isolated specimens of
colour to males) display a bright milky white band on Polythore concinna. Field work was done during the
the wings while the other phenotype of females (gy- dry season, between 30 October and 16 December
nochrome, different colour to males) shows a bright 2014 (see Sanmartín–Villar and Cordero–Rivera,
orange band in the forewing and a violet band on 2016, for further details).
the hindwing (Sanmartín–Villar and Cordero–Rivera, In both localities, one to two observers walked over
2016). Beccaloni (1997) described several mimicry small streams and forest paths where the damselflies
rings for the Ithomiinae butterflies in Jatun Sacha were found, for an average of 7 h per day. The animals
Animal Biodiversity and Conservation 42.2 (2019) 295

Fig. 1. A P. mutata male (A) from TBS and a P. derivata male (marked) (B) from JSBS. The insert in (A)
shows the tip of the abdomen of a male P. mutata with algae growing, which is suggestive of old age.
Pictures by ACR.

Fig. 1. Un macho de P. mutata (A) de la estación biológica de Tiputini (TBS) y uno marcado de P. derivata
(B) de la estación biológica de Jatun Sacha (JSBS). El círculo en (A) muestra el extremo del abdomen
de un macho de P. mutata con algas, lo que sugiere que es de edad avanzada. Fotografias de ACR.

were captured with an entomological net, measured populations (the linear distance is 167.2 km, estimated
with a calliper (body length to the nearest 0.1 mm, from GoogleEarth). We used the Information Theory
including anal appendages), marked with a white Approach to rank models by their Akaike’s Information
number in the wing (xylene–free white permanent Criterium (AIC), which is minimised in the models
marker; Pilot Super Color EF: www.pilotpen.com; that are better supported by the data (Burnham and
fig.  1), photographed, and released. Re–sightings Anderson, 1998). However, this method does not
were carried out with the naked eye or using a Minox allow to know whether the candidate models are good
monocular, so that animals were only netted again if enough to explain the variability of the data. There-
needed for specific reasons. fore, we first tested the fit of the full time–dependent
Cormack–Jolly–Seber model by groups using program
Mark and recapture analyses Release. The model is defined by:
Phi(g*t) p(g*t)
To analyse mark–recapture histories we used the
software Mark 8.1 (White and Burnham, 1999). Our where Phi and p represent the recapture and survival
analyses included sex and time (and their interaction) probability, g is sex and t is time.
as factors to test for their effects on survival and re- In the case of TBS, this model showed good
capture probabilities. Each population was analysed adjustment for males of P. mutata (Goodness of fit
separately, because migration is extremely unlikely results (Test 2 + Test 3): x218 = 15.71, p = 0.613),
due to the high geographic distance between the but due to the low recapture rate of females, this
296 Cordero–Rivera et al.

1,0

0,9

0,8

Survival probability 0,7

0,6

0,5

0,4

0,3

0,2

0,1

0,0
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
Body length (mm)

Fig. 2. The relationship between body length and survival probability of P. mutata at Tiputini (sexes
combined). Note that the estimates suggest stabilizing selection. Grey lines indicate 95 % confidence
intervals.

Fig. 2. Relación entre la longitud corporal y la tasa de supervivencia de P.  mutata en Tiputini (sexos
combinados). Nótese que las estimaciones sugieren la existencia de una selección estabilizadora. Las
líneas grises indican el intervalo de confianza del 95 %.

test could not be calculated. The adjustment was of parameters were obtained by model averaging
clear in the case of P. mutata from JSBS (Test 2 + 3 using Mark software and are shown as average pro-
by groups: males x252 = 22.73, p = 1; females bability ± SE. We estimated the expected longevity
x227  = 8.21, p = 1). For P. derivata at JSBS, only using the formula of Cook et al. (1967):
males were recaptured, and again the saturated
model showed good adjustment (Test 2 + 3: males Longevity = 1 / loge(survival)
x214 = 9.10, p = 0.825). In JSBS we ran an analysis
of P. mutata females only to test for differences
between morphs. The saturated model was also Results
appropriate (Test 2 + 3 by morph: androchrome
x27 = 0.00, p = 1; gynochrome x27 = 9.45, p = 0.977). Overall, population density of both Polythore species
The second step was to estimate the variance was low. At TBS, we marked 21 males and 13 females
inflation parameter (c–hat) by two methods. We divi- of P. mutata in December 2012 and 16 males and
ded the c–hat obtained from model Phi(g*t) p(g*t) by 5 females in February 2013, and recaptured four males
the mean c–hat of the bootstrap simulations in Mark. and one female from 2012. No marked animals were
With this method we always obtained c–hat = 1. The found in June 2013. At JSBS we marked 76 males
second method divided the deviance of the saturated and 35 females of P. mutata, 19 males and five fe-
model by the mean deviance of the bootstrap pro- males of P. derivata and three males and one female
cedure. Values ranged between 0.98 and 1.27. No of P. concinna. This last species is not analysed here
correction was done when c–hat was lower than one, due to the lack of recaptures. Female polymorphism
but otherwise c–hat estimates were used to adjust was detected in both populations of P. mutata. In TBS
parameter estimates and standard errors. we only observed one androchrome female in the first
In the next step, we ran models without individual period (that could not be marked) and marked another
covariates, to select the most supported model, as androchrome female in the second sampling period,
the one that minimizes Quasi Akaike's Information with an overall frequency of 5.6 % androchromes (1
Criterion (QAIC). Once these models were identified, out of 18 females), while at JSBS, androchromes were
body length was included as an individual covariate, 40 % (14 out of 35). No female polymorphism was
both as linear and quadratic terms. When there was detected in P. derivata (see also Sanmartín–Villar and
statistical support for more than one model, estimates Cordero–Rivera, 2016).
Animal Biodiversity and Conservation 42.2 (2019) 297

A 1,0

0,9

0,8
Survival probability 0,7

0,6

0,5

0,4

0,3

0,2

0,1

0,0
37 38 39 40 41 42 43 44 45 46 47 48
Body length (mm)
B 1,0

0,9

0,8
Survival probability

0,7

0,6

0,5

0,4

0,3

0,2

0,1

0,0
37 38 39 40 41 42 43 44 45 46 47 48
Body length (mm)

Fig. 3. The relationship between body length and survival probability of P. mutata at Jatun Sacha.
There was positive directional selection in both sexes: A, males; B, females. (Grey lines indicate 95%
confidence intervals).

Fig. 3. Relación entre la longitud corporal y la probabilidad de supervivencia de P. mutata en Jatun

Sacha. Hubo selección direccional positiva en ambos sexos: A, machos; B, hembras. (Las líneas grises
indican el intervalo de confianza del 95 %).

Considering both sexes, at TBS model selection term, suggesting stabilizing selection (fig. 2). Given
by AICc indicates that the best model is {Phi(.) that the best model has a D AICc  = 3.8 (table 1s),
p(g)}, which means similar survival for males and no model averaging was needed, and we esti-
females and a different recapture rate by sex. The mated daily survival rate from the best model as
inclusion of body size as a covariate (table 1s) 0.974  ±  0.011 (expected longevity 38 days) and
indicates that survival is affected by body size, recapture rates 0.391 ± 0.047 for males and 0.081
and that the best model includes the quadratic ± 0.041 for females.
298 Cordero–Rivera et al.

At JSBS, the model selected by AICc is {Phi(g) survival rely on individually marked animals, for which
p(.)}, the opposite to the previous case (table 2s), the exact time of death is most often unknown, be-
indicating higher survival rates for females, while the cause marked individuals cannot be followed closely
recapture rate was similar for males and females. through time. These limitations are to be taken into
Including body length as a linear covariate improved account when interpreting our results.
the fit of this model, but the quadratic term diminished For other Polythoridae, like Polythore procera,
fit. Therefore, our data for JSBS suggest a positive estimates of survival are ~0.97 (Sánchez–Herrera
directional selection for larger body length (fig. 3) in and Realpe, 2010), and ~0.40 to 1, depending on the
both sexes. Using model averaging, the daily survi- population for P. gigantea (Altamiranda–Saavedra and
val rate was estimated as 0.821 ± 0.032 for males Ortega, 2012), which are similar to our estimates. The
(expected longevity 5.1 days) and 0.953 ± 0.019 for analysis of recapture rates suggests that females were
females (20.8 days), with a common recapture rate less often resighted at TBS, whereas at JSBS, about
of 0.200 ± 0.024. 20% of individuals were resighted, irrespective of sex.
The analysis of recapture histories of the JSBS In contrast, recapture probabilities for the Andean
polymorphic females of P. mutata was limited by the species previously studied yielded higher estimates
low number of specimens found. On some dates, (Sánchez–Herrera and Realpe, 2010; Altamiranda–
only one morph was found, and therefore interactions Saavedra and Ortega, 2012) than for our Amazonian
morph*time could not be calculated at these dates, and species. Andean species of Polythore showed higher
models including these interactions were excluded. The density of individuals overall than those in the Ama-
best model, as selected by AICc, was {Phi(.) p(g)}, and zonian field sites we evaluated here. The preferred
the second {Phi(g) p(g)}, and both were much more habitat of Polythore (i.e. rocky waterfalls, small creeks)
supported than the next models (DAICc  > 4.3; table is not common in the Amazon in comparison with the
2s). Using model averaging, we found the daily sur- Andean foothills. In particular, the difference between
vival rate was estimated as 0.927 ± 0.034 (expected the TBS and JSBS may be explained by the fact that
longevity of 13.2 days) for androchromes and 0.943 observations at TBS were mainly done near the stream,
± 0.021 (longevity 17.0 days) for gynochromes, and where males showed high site fidelity and territorial
recapture rates 0.088 ± 0.038 for androchromes and behaviour, whereas at JSBS the topography allowed
0.247 ± 0.044 for gynochromes. Recapture rates were a more extensive examination of the areas around
clearly different between morphs, and there is some the streams and in the forest. In fact, some females
evidence for higher survival by gynochromes. were found feeding at the same sunspots in the forest
Finally, we analysed the recapture histories of for several days. Therefore, our data suggest that the
males of P. derivata from JSBS (no females were lower recapture rate of female damselflies (Cordero–
recaptured). In this case, given the small sample size Rivera and Stoks, 2008) is due to females remaining
(N = 16 males) we could not use covariates. The best at larger distances from the water. This behaviour was
model was the simplest one, with constant survival also observed for the Andean P. procera, where the
and recapture probabilities ({Phi(.) p(.)} indicating a females seem to remain at highly dense forest areas,
survival rate of 0.962 ± 0.018 and a recapture rate of while males remained near the open stream area
0.241 ± 0.048, with an expected longevity of 25.6 days. (Sánchez–Herrera et al., 2010).
Some observations might suggest higher lifespan At TBS, we found that survival probabilities were
than our estimates. For instance, some individuals not affected by sex, while at JSBS they were higher
showed algae and even bryophytes over the body for females, and were in the interval 0.821–0.974.
surface (fig. 1A) at TBS. In addition, five individuals These survival values translate into longevities of
marked in December at TBS, were found alive in Fe- five to 38 days, similar to the maximum longevity
bruary, with a lifespan between 54 and 63 days. The observed in other damselflies from temperate regions
proportion of animals with algae on the abdomen was (Cordero–Rivera and Stoks, 2008). Therefore, we did
15.6 % in December and 38.5 % in February, when not find evidence of long pre–reproductive periods or
most of the specimens were clearly older. Three of high survival, contrary to our expectations. However,
the animals surviving from December to February the presence of algae and liverworts growing on the
showed algae on the abdomen, but none of these abdomen and wings of some specimens (fig. 1A)
had algae present when first marked. suggests that they have lived in the humid forest for
months (Lücking et al., 2010) because these algae
cannot be acquired during the larval stage (the cuticle
Discussion changes after metamorphosis) and because liverworts
have a long life cycle (Lücking et al., 2010). Our
We found that Polythore mutata and P. derivata dam- estimates of maximum lifespan are thus likely below
selflies have a high survival rate and low recapture the real values. To test this hypothesis, future studies
probabilities, particularly in females. Mark–recapture should include longer periods of fieldwork to maximize
studies provide information based on the subset of the probability of detecting a particularly high lifespan.
animals that are recaptured, and therefore the esti- Records of algae growing on rainforest odonates are
mates are valid only if temporal emigration does not very rare and have been interpreted as indicators
occur, and when emigration is permanent, this cannot of old age (e.g. Fincke and Hadrys, 2001). Algae
be disentangled from mortality (Lebreton et al., 1992). were also found in other Ecuatorian species such as
Examination of trade–offs between reproduction and Metaleptobasis sp, Argia oculata, Heteragrion cooki,
Animal Biodiversity and Conservation 42.2 (2019) 299

Hetaerina fuscoguttata, and Uracis imbuta (ACR, ISV adults and longer larval life (Waller and Svensson,
and ART personal observations, 2014, 2016), and 2017). Survivorship seems more related to body size
some odonates in rainforests in Yunnan (China, ACR, for JSBS males than for females (compare the slopes
pers. obs. 2016). However, even the longest–living in fig. 3). This could be explained by the territorial
specimens in our samples were unlikely to live for more behaviour and the positive relationship between body
than two months. Higher longevity could also be the size and reproductive success in males and the lack of
result of the low activity levels of Polythore individuals, relationship for this trait and fitness in many females
which remained perched immobile almost all the time (Serrano–Meneses et al., 2007; Sanmartín–Villar
(see Sanmartín–Villar and Cordero–Rivera, 2016), but and Cordero–Rivera, 2016). In the case of TBS our
nevertheless showed nervous behaviour (they quickly results suggest stabilizing selection (fig. 2). It is pos-
escaped when a human approached) which is probably sible that small males were excluded from the best
an efficient antipredator strategy. Finally, their possible areas by the aggressive behaviour of larger males.
mimicry to poisonous butterflies could also contribute However, survivorship also decreased in the largest
to high survival (Beccaloni, 1997). This topic is worth body sizes. This might be due to the associated
exploring in future field experiments. cost of large bodies (e.g. maintenance, decrease of
Female colour polymorphism for P. mutata was also manoeuvrability, conspicuousness towards predators)
evidenced in the TBS population, which constitutes or/and the different directions of natural and sexual
the first evidence of this polymorphism out of JSBS. selection in those traits (Outomuro et al., 2016b). Our
However, we only saw two andromorphs throughout results indicate that survival can or cannot be posi-
the whole sampling dates for TBS. Therefore, even tively related to larger body size in these damselflies,
though there are polymorphic individuals in different possibly depending on local ecological conditions. A
populations, the selective forces maintaining the po- previous study of another forest damselfly found that
lymorphism might differ depending on the geographic in a sunny stream, body size was negatively correlated
location. Beccaloni's (1997) suggestion of convergence with survival, but in a shaded environment it was not
of these morphs within two Ithomiinae mimicry rings for (Rivas–Torres et al., 2017). Another possibility is that
JSBS could play a role as a selective force potentially survival may show interannual variation, a suggestion
maintaining higher survival probabilities for P. mutata. that also merits further study. It seems likely that
Our results show little evidence for an effect of female contrasting selective regimes between natural and
colour morph on survival rate in females of P. mutata. sexual selection in adults, between larvae and adults
These results are to be taking with caution because the (Waller and Svensson, 2017), and between years and
number of females resighted was low, but they agree localities contributes to the stasis observed on body
with previous work on other polymorphic damselflies size over time.
(e.g. Andrés and Cordero–Rivera, 2001) that suggest
no effect of female colour morphs on survival. Few
studies however have applied the modern techniques Acknowledgements
of capture–recapture analysis (Lebreton et al., 1992) to
polymorphic damselflies, and further research is clearly Funding was provided by a grant from the Spanish
needed on this topic. The conspicuousness of the wings Ministry of Economy and Competitiveness, including
in male and androchrome females of P. mutata might FEDER funds (CGL2014–53140–P). ISV and ART
be higher than that in gynochrome females, at least were supported by FPI grants (BES–2012–052005
for predators that perceive the UV range (Bick and and BES–2015–071965).
Bick, 1965). In addition, no androchrome females were
observed near the streams where males concentrated,
even in JSBS, where androchromes were common. References
This suggests the two female phenotypes have diffe-
rent reproductive strategies, a common phenomenon Altamiranda–Saavedra, M., Ortega O., 2012. Estruc-
in other damselflies (Van Gossum et al., 2008). Future tura poblacional de Polythore gigantea (Odonata:
experiments are needed to decipher the significance Polythoridae) en sistemas lóticos con diferentes
of this polymorphism. estados de conservación. Revista de Biología
We found that adding body length as a covariate Tropical, 60: 1205–1216.
in the mark–recapture analysis increases the fit of Andrés, J. A., Cordero–Rivera, A., 2001. Survival rates
the models. At TBS, we found evidence that suggests in a natural population of the damselfly Ceriagrion
stabilizing selection for body size, whereas at JSBS tenellum: Effects of sex and female phenotype.
larger body length was apparently under positive Ecological Entomology, 26: 341–346.
selection. In fact, the average size of males at JSBS Beccaloni, G. W., 1997. Ecology, natural history and
(42.9 ± 0.27, SE, N = 72) was smaller than the size behaviour of ithomiine butterflies and their mimics
of males at TBS (43.9 ± 0.45, N = 34; t104  = –2.109, in Ecuador (Lepidoptera: Nymphalidae: Ithomiinae).
p = 0.037). A recent review of body size evolution in Tropical Lepidoptera, 8: 103–124.
the Odonata found a consistent trend of positive selec- Bick, G. H., Bick, J. C., 1965. Color variation and
tion for larger size, but at the same time no evidence significance of color in reproduction in the damselfly
for an increase in body size over evolutionary time. Argia apicalis (Say) (Zygoptera: Coenagriidae).
The authors suggested that this lack of concordance Canadian Entomologist, 97: 32–41.
is due to the trade–off between larger body size as Burnham, K. P., Anderson, D. R., 1998. Model selec-
300 Cordero–Rivera et al.

tion and inference. A practical information–theoretic res, M., Cordero–Rivera, A., 2017. Demographics
approach. Springer, New York. and behaviour of Heteragrion cooki, a forest
Cook, L. M., Brower, P. P., Crozer, H. J., 1967. The damselfly endemic to Ecuador (Odonata). Interna-
accuracy of a population estimation from multiple re- tional Journal of Odonatology, 20: 123–135, Doi:
capture data. Journal of Animal Ecology, 36: 57–60. 10.1080/13887890.2017.1336495
Cordero–Rivera, A., Stoks, R., 2008. Mark–recaptu- Sánchez–Herrera, M., Realpe, E., 2010. Population
re studies and demography. In: Dragonflies and structure of Polythore procera at a Colombian
damselflies. Model organisms for ecological and stream (Odonata: Polythoridae). International
evolutionary research: 7–20 (A. Córdoba–Agui- Journal of Odonatology, 13: 27–37.
lar, Ed.). Oxford University Press, Oxford, Doi: Sánchez–Herrera, M., Realpe, E., Salazar, C.,
10.1093/acprof:oso/9780199230693.003.0002 2010. A neotropical polymorphic damselfly shows
Cordero Rivera, A., Andrés Abad, J. A., 1999. Lifetime poor congruence between genetic and traditional
mating success, survivorship and synchronized morphological characters in Odonata. Molecular
reproduction in the damselfly Ischnura pumilio Phylogenetics and Evolution, 57: 912–917.
(Odonata: Coenagrionidae). International Journal Sanchez Herrera, M., Beatty, C., Nunes, R., Realpe,E.,
of Odonatology, 2:105–114. Salazar, C., Ware, J. L., 2018. A molecular syste-
Córdoba–Aguilar, A., Cordero–Rivera, A., 2005. Evo- matic analysis of the Neotropical banner winged
lution and ecology of Calopterygidae (Zygoptera: damselflies (Polythoridae: Odonata). Systematic
Odonata): status of knowledge and research pers- Entomology, 43: 56–67, Doi: 10.1111/syen.12249
pectives. Neotropical Entomology, 34: 861–879, Sánchez Herrera, M., Kuhn, W. R., Lorenzo–Carballa,
Doi: 10.1590/S1519–566X2005000600001 M. O., Harding, K. M., Ankrom, N., Sherratt, T.
Cornell, H. V., Hawkins, B. A., 1995. Survival patterns N., Hoffmann, J., Van Gossum, H., Ware, J. L.,
and mortality sources of herbivorous insects: some Cordero–Rivera, A., Beatty, C. D., 2015. Mixed
demographic trends. The American Naturalist, signals? Morphological and molecular evidence
145: 563–593. suggest a color polymorphism in some Neotropical
Fincke, O. M., Hadrys, H., 2001. Unpredictable offs- Polythore damselflies. PLOS One, 10: e0125074,
pring survivorship in the damselfly, Megaloprepus Doi: 10.1371/journal.pone.0125074
coerulatus, shapes parental behavior, constrains Sanmartín–Villar, I., Cordero–Rivera, A., 2016. Fema-
sexual selection, and challenges traditional fitness le colour polymorphism and unique reproductive
estimates. Evolution, 55: 762–772. behaviour in Polythore damselflies (Zygoptera: Po-
Van Gossum, H., Sherratt, T. N., Cordero–Rivera, lythoridae). Neotropical Entomology, 45: 658–664,
A., 2008. The evolution of sex–limited colour Doi: 10.1007/s13744–016–0417–7
polymorphisms. In: Dragonflies and damselflies. Serrano–Meneses, M. A., Córdoba–Aguilar, A., Mén-
Model organisms for ecological and evolutionary dez, V., Layen, S. J., Székely, T., 2007. Sexual
research: 219–229 (A. Córdoba–Aguilar, Ed.). size dimorphism in the American rubyspot: male
Oxford University Press, Oxford. body size predicts male competition and mating
Lebreton, J. D., Burnham, K. P., Clobert, J., Ander- success. Animal Behaviour, 73: 987–997.
son, D. R., 1992. Modeling survival and testing Sherratt, T. N., Laird, R. A., Hassall, C., Lowe, C.
biological hypotheses using marked animals: a D., Harvey, I. F., Watts, P. C., Cordero–Rivera,
unified approach with case studies. Ecological A., Thompson, D. J., 2010. Empirical evidence of
Monographs, 62: 67–118. senescence in adult damselflies (Odonata: Zygop-
Lücking, R., Mata–Lorenzen, J., Dauphin, L., 2010. tera). Journal of Animal Ecology, 79: 1034–1044.
Epizoic liverworts, lichens and fungi growing on Tauber M. J., Tauber, C. A., 1976. Insect seasonality:
Costa Rican Shield Mantis (Mantodea: Choera- diapause maintenance, termination, and postd-
dodis). Studies on Neotropical Fauna and Envi- iapause development. Annu. Rev. Entomol., 21:
ronment, 45: 175–186. 81–107.
Outomuro, D., Ángel–Giraldo, P., Corral–Lopez, A., Waller, J. T., Svensson, E. I., 2017. Body size evolution
Realpe, E., 2016a. Multitrait aposematic signal in in an old insect order: no evidence for Cope’s Rule
Batesian mimicry. Evolution, 70: 1596–1608, Doi: in spite of fitness benefits of large size. Evolution,
10.1111/evo.12963 71: 2178–2193, Doi: 10.1111/evo.13302
Outomuro, D., Söderquist, L., Nilsson–Örtman, V., White, G. C., Burnham, K. P., 1999. Program MARK:
Cortázar–Chinarro, M., Lundgren, C., Johansson, Survival estimation from populations of marked
F., 2016b. Antagonistic natural and sexual selection animals. Bird Study, 46 Supplem.: 120–138.
on wing shape in a scrambling damselfly. Evolution, Wikelski, M., Hau, M., Wingfield, J. C., 2000.
70: 1582–1595, Doi: 10.1111/evo.12951 Seasonality of reproduction in a neotropical rain
Rivas–Torres, A., Sanmartín–Villar, I., Gabela–Flo- forest bird. Ecology, 81: 2458–2472.
Animal Biodiversity and Conservation 42.2 (2019) I

Table 1s. Results of model selection for P. mutata at TBS. Survival rate is denoted by Phi and recapture
probability by p. The group is the sex (g) and time variation is indicated by t. The best models include
body length as covariate (body).

Tabla 1s. Resultados de la selección de modelos para P.  mutata en la estación biológica de Tiputini
(TBS). La tasa de supervivencia se indica como Phi y la probabilidad de recaptura, como p. El grupo es
el sex (g) y la variación temporal se indica como t. Los mejores modelos incluyen la longitud corporal
como covariable (body).

AICc Model Num.

Model AICc
DAICc Weights Likelihood Par Deviance
{Phi(body+body2) p(g)} 263.923 0.000 0.787 1.000 5 253.292
{Phi(body) p(g)} 267.735 3.812 0.117 0.149 4 259.319
{Phi(.) p(g)} 269.092 5.169 0.059 0.075 3 262.845
{Phi(g) p(g)} 270.343 6.419 0.032 0.040 4 261.926
{Phi(.) p(g*t)} 275.353 11.429 0.003 0.003 23 215.015
{Phi(.) p(t)} 277.137 13.213 0.001 0.001 13 246.953
{Phi(g) p(t)} 278.053 14.129 0.001 0.001 14 245.169
{Phi(g) p(g*t)} 278.803 14.880 0.000 0.001 24 215.014
{Phi(t) p(g)} 281.304 17.381 0.000 0.000 14 248.421
{Phi(t) p(t)} 293.578 29.654 0 0 24 229.788
{Phi(g) p(.)} 293.850 29.927 0 0 3 287.603
{Phi(.) p(.)} 294.091 30.168 0 0 2 289.969
{Phi(t) p(g*t)} 300.979 37.056 0 0 34 196.919
{Phi(t) p(.)} 301.301 37.378 0 0 13 271.117
{Phi(g*t) p(g)} 308.796 44.872 0 0 24 245.006
{Phi(g*t) p(.)} 321.948 58.025 0 0 23 261.611
{Phi(g*t) p(t)} 324.809 60.885 0 0 34 220.748
{Phi(g*t) p(g*t)} 330.635 66.711 0 0 40 195.968
II Cordero–Rivera et al.

Table 2s. Results of model selection for P. mutata at JSBS. Survival rate is denoted by Phi and
recapture probability by p. The group is sex (g) and time variation is indicated by t. The best models
include body length as covariate (body).

Tabla 2s. Resultados de la selección de modelos para P. mutata en la estación biológica de Jatun Sacha
(JSBS). La tasa de supervivencia se indica como Phi y la probabilidad de recaptura, como p. El grupo
es el sex (g) y la variación temporal se indica como t. Los mejores modelos incluyen la longitud corporal
como covariable (body).

AICc Model Num.

Model QAICc
DQAICc Weights Likelihood Par QDeviance
{Phi(g+body) p(.)}–not standardized}
534.124 0.000 0.748 1.000 4 525.917
{Phi(g+body+body2) p(.)}–not standardized}
536.691 2.567 0.207 0.277 6 524.253
{Phi(g) p(.)} 541.009 6.886 0.024 0.032 3 534.886
{Phi(g) p(g)} 542.985 8.861 0.009 0.012 4 534.779
{Phi(.) p(.)} 543.619 9.495 0.006 0.009 2 539.558
{Phi(g+body) p(.)}–standardized}
544.878 10.754 0.003 0.005 4 536.672
{Phi(.) p(g)} 545.573 11.449 0.002 0.003 3 539.450
{Phi(g) p(t)} 595.728 61.604 0 0 41 491.791
{Phi(.) p(t)} 596.340 62.217 0 0 40 495.581
{Phi(t) p(.)} 610.211 76.087 0 0 39 512.588
{Phi(t) p(g)} 613.337 79.213 0 0 40 512.578
{Phi(.) p(g*t)} 710.521 176.398 0 0 76 462.587
{Phi(g) p(g*t)} 711.179 177.055 0 0 77 457.906
{Phi(t) p(t)} 712.677 178.554 0 0 74 475.161
{Phi(g*t) p(.)} 724.970 190.846 0 0 73 492.538
{Phi(g*t) p(g)} 730.035 195.912 0 0 74 492.519
{Phi(g*t) p(t)} 924.270 390.146 0 0 107 456.292
{Phi(t) p(g*t)} 938.379 404.256 0 0 110 440.879
{Phi(g*t) p(g*t)} 1319.746 785.622 0 0 137 425.877