Vis Egg 2022
Vis Egg 2022
Vis Egg 2022
https://doi.org/10.1007/s10695-020-00844-2
Received: 15 January 2020 / Accepted: 22 June 2020 / Published online: 30 June 2020
# Springer Nature B.V. 2020
Abstract Assessing female fish reproductive success were hydrated in water. After 24 h, a picture of the eggs
requires a thorough evaluation of egg characteristics, was obtained using a dedicated shooting system
including egg number, size, and variability as well as consisting of a light source and a digital single-lens
egg developmental potential through the monitoring of reflex (SLR) camera. An image processing algorithm
embryo survival after fertilization. While embryonic was developed to allow the automatic detection and
success relies, at least in part, on paternal contribution, separation of the eggs and to perform automatic mea-
some parameters are strictly related to egg characteris- surements of egg number and individual egg size. The
tics, one of the main ones being the viability of the egg presence of white egg was used as an indirect measure
when released into the water at spawning. It is however of egg integrity, the “whitening” being the result of
not necessarily possible, at least in salmonid fish that lay water entry into the egg through the vitelline membrane.
nontransparent eggs, to separate the different causes of These white eggs were therefore considered nonviable,
egg/embryo failure. In this context, our aim was (i) to as a result of their lack of physical integrity. Fertilization
develop a simple and rapid system to capture images of assays were performed in parallel using a subsample of
rainbow trout eggs combined with computerized pro- the same egg batch. Embryonic development was mon-
cessing of these images to perform a fully automatic itored and hatching rate was calculated. A significant
individual characterization of egg features including correlation between white egg percentage after hydra-
number and size (ii) to estimate unfertilized egg viability tion and hatching rate was observed (Spearman coeffi-
through the monitoring of the percentage of eggs that cient = −0.557, p < 0.001), in consistency with the fact
will not survive to water hydration. To evaluate the that nonviable egg will not allow successful embryonic
VisEgg system, unfertilized eggs (approximatively 400 development. In contrast, the percentage of eggs that do
eggs per batch) originating from 105 different females not successfully hatch includes egg/embryo failures of
different nature including reduced egg viability. Using
the VisEgg, we were able to quantify the lack of viabil-
Emilie Cardona and Jerome Bugeon contributed equally to this ity of the eggs separately from the different other events
work.
that may occur during fertilization and incubation. the
Electronic supplementary material The online version of this VisEgg is a convenient and reliable tool to obtain indi-
article (https://doi.org/10.1007/s10695-020-00844-2) contains vidual measures on trout eggs. It can be used to assess
supplementary material, which is available to authorized users.
not only egg size and egg number but also unfertilized
E. Cardona : J. Bugeon : E. Segret : J. Bobe (*) egg viability before fertilization.
INRAE, LPGP, F-35000, Rennes, France
evaluation of developmental success that is performed Flavobacterium infection. Two days after the detection
on fertilized eggs to separate the origin of egg quality of ovulation, females were manually stripped. Eggs
defects (viable eggs vs. nonviable eggs that do not allow were weighted and two samples were taken for fertili-
developmental success). zation assays and for VisEgg phenotyping analysis. A
total of 105 spawns were individually analyzed.
Ethical statements For each spawn (i.e., each female, n = 105), approxi-
mately 400 eggs were fertilized with a pool of sperm
Experimentations were conducted in the INRAE collected from males fed ad libitum on a commercial
PEIMA experimental facility (Sizun, France— diet and embryonic success was monitored. Fifteen
Agreement number B29-277-02). All fish were reared microliters of a pool of semen obtained from five males
and handled in strict accordance with French and Euro- presenting the highest sperm motility were used, and 15
pean policies and guidelines of the INRAE PEIMA mL of Actifish solution (sperm motility activating saline
Institutional Animal Care and Use Ethical Committee, solution, IMV technologies, L’Aigle, France) were
which specifically approved this study. Fish were mon- added onto the eggs. Five minutes later, the sperm
itored daily during the experiment. If any clinical symp- motility activating solution was drained and egg batches
toms (i.e., morphological abnormality, restlessness, or were transferred into individual incubators in a
uncoordinated movements) were observed, fish were recirculated water unit. Water temperature (12.0–12.6
sedated by immersion in MS-222 solution at a concen- °C) was monitored daily. Dead eggs and embryos were
tration of 50 mg L−1 and then euthanized by immersion periodically manually counted and removed. Survival at
in a MS-222 solution at a concentration of 400 mg L−1 eyeing, hatching, and completion of yolk-sac resorption
(anesthetic overdose) in 3 min. (YSR) were monitored and calculated as a percentage of
the initial number of eggs used for fertilization. The
Broodstock breeding and experimental design occurrence of noticeable morphological malformations
(spinal cord torsion, head or caudal fin malformations,
Female rainbow trouts from an autumn-spawning strain etc.) at YSR was also recorded (Supplementary
were held under natural photoperiod until their first Table 1).
reproduction (2 years) in the INRAE PEIMA experi-
mental facility. After spawning, fish (initial weight: Phenotyping analysis
1693 ± 306 g) were transferred into 6 outdoor 2-m3
tanks. Fish were fed on a commercial trout broodstock For VisEgg phenotyping, unfertilized eggs (20 to 25 g
diet (Le Gouessant, Lamballe, France) for a period of 5 corresponding to approximatively 400 eggs) were stored
months. During this 5-month period, an artificial photo- in a 100-mL container and then water from the incuba-
period regime was applied to obtain a second reproduc- tion system was added to hydrate the eggs (80 mL).
tion during summer: a 2.5-month-long photoperiod After a 24-h hydration at 4 °C, a picture of the egg
(20 h light, 4 h dark from December to March 2017) sample was taken using a dedicated commercially avail-
followed by a 2.5-month-short photoperiod (8 h light, able shooting system. This system consisted in a light
16 h dark from March to June 2017). The water tem- tablet and a digital single-lens reflex (SLR) camera
perature during the experiment ranged from 6 to 12 °C. (canon EOS 1000D, resolution: 10.1 M pixels) (Fig.
1). Eggs were rinsed with water and delicately poured
Sample acquisition into a petri dish (diameter = 14 cm). The image calibra-
tion was performed with an image of a 1-euro coin with
During spawning season (June 21 to August 3, 2017), a known area to calculate the pixel size. The petri dish
females were checked for ovulation once a week by was then placed on the light tablet. It was verified that
applying a manual pressure onto the abdomen. When the eggs were not superimposed and a picture was taken.
ovulation was detected, fish were given an intraperito- An image processing algorithm was developed and
neal injection of an antibiotic (Vetrimoxin) to prevent coded using a Visual Basic for Applications (VBA)
674 Fish Physiol Biochem (2021) 47:671–679
macro with Visilog 7.3 software (Thermo Scientific). analysis was thus validated. The Spearman coefficients
This software allows the automatic detection and sepa- were calculated. For statistical analysis performed to
ration of the eggs in order to make fully automatic compare the two calculating methods of hatching rate
measurements of egg number and specific measurement (using all eggs or on only viable eggs, see Fig. 5), a
for each egg (size, shape, color, white egg occurrence). Kruskal-Wallis test was used followed by Dunn test’s
The total number of ovulated eggs per females could be post hoc analysis.
back calculated using the number of measured eggs and
the total weight of the spawn. The workflow of the
VisEgg macro-command is presented in Fig. 2. All
details and relevant information about the VisEgg image Results and discussion
analysis workflow are available here: https://github.
com/LpgpImage/VisEgg/wiki. Automatic assessment of egg features
The egg size remains an interesting parameter to individual measures such as egg size and intra-spawn
measure. Although the possible benefit effect of egg size variability.
size on egg survival and alevins is still debated, espe-
cially under aquaculture conditions (Bobe and Labbé Assessment of egg viability regardless of fertilization
2010; Bromage et al. 1992; Campbell et al. 1992; and subsequent developmental success
Jastrebski and Morbey 2009; Migaud et al. 2013;
Springate and Bromage 1985b), it is established that Unfertilized eggs rather than fertilized eggs were used to
larger eggs produce larger alevins (Springate and evaluate our phenotyping tool because the hardening
Bromage, 1985) with substantial fitness advantages process is independent of fertilization (Lahnsteiner et al.
over small alevins (Heath et al. 2003). While egg size 1999). Two distinct phenotypes can be observed after the
is very often measured in studies on fish reproductive hydration process: either the egg hydrates normally and
performance, it is unusual to measure egg size vari- grows slightly or the egg turns white. This phenotype is
ability within the spawn. This is certainly due to the easily observable on images (Fig. 3D). Moreover, the
difficulty of assessing this parameter. Female rainbow VisEgg can separate normal (i.e., viable) eggs from white
trout (Oncorhynchus mykiss) is a group-synchronous (i.e., nonviable) eggs and then calculate the percentage of
species which produces a single spawn each year white egg for each analyzed egg batch (Fig. 3C).
where all oocytes develop and ovulate at the same In this study, multiple linear regression was per-
time (Lubzens et al. 2010). In normal conditions, eggs formed to test the correlation between hatching rate
from a single spawn are of homogeneous size; the and the different parameters measured using the VisEgg
presence of large egg size intra-spawn variability in (i.e., egg diameter, variability of egg diameter, and white
salmonid can be the consequence of a disruption of egg percentage). Only white egg percentage was signif-
the physiological processes underlying oogenesis icantly correlated with hatching rate. In contrast, egg
(Jastrebski and Morbey 2009). The measurement of diameter and variability of egg diameter appear to have
this parameter therefore appears to be relevant for no link with egg quality and survival of alevins (Spear-
rainbow trout reproduction studies. Thus, the VisEgg man coefficient = 0.39 and −0.30, respectively, for egg
is a convenient and reliable tool to obtain accurate diameter and variability of egg diameter). Linear
676 Fish Physiol Biochem (2021) 47:671–679
Fig. 3 Images of trout eggs as processed by the VisEgg. (A) processed by the VisEgg for an automatic detection of white eggs
Initial image obtained with the VisEgg system equipment present- and normal eggs (white eggs and normal eggs are represented in
ed in Fig. 1. (B) Image as processed by VisEgg for an automatic red and blue colors, respectively). (D) Pictures with macro objec-
detection and separation of the eggs in order to make fully auto- tive illustrating normal and white eggs after hydration
matic measurements of egg number and size. (C) Image as
regression between white egg percentage and hatching and to develop into an embryo (or to die at different
rate shows a significant negative correlation (p-value = times during development). Figure 4 illustrates the suc-
2.6 x10-10; Spearman coefficient = −0.56). It should be cessive different origins of egg/embryo failure that can
noted that we observed highly significant correlations occur between spawning and yolk-sac resorption. The
between surviving rates at eyeing and hatching as well graph shows the mean values obtained using the 105
as between surviving rates at hatching and yolk-sac different spawns analyzed in the present study.
resorption (YSR) (p-value = < 2 × 10-16, Spearman Figure 5A shows the hatching rate calculated using all
coefficient = −0.99 for both correlations). Surviving eggs (in blue) or only using viable eggs (in red), the
rates at eyeing and YSR are therefore also significantly batches being ranked based on the overall hatching rate
negatively correlated with the percentage of nonviable (blue dots). The blue and red dots do not follow exactly
eggs (p-value = 9.9 × 10-11, Spearman coefficient = the same pattern due to the differences in the origins of
−0.58 and p-value = 4.7 x10−10, Spearman coefficient egg/embryo failure up to hatching, with major differences
= −0.55 for eyeing and YSR, respectively). between the two groups for some individuals. In some
The negative correlation between white egg percent- cases, when the red and blue dots are far apart, there is a
age and developmental success was expected because high incidence of egg viability on the overall develop-
nonviable eggs will not allow successful embryonic mental success. In contrast, when the red and blue dots
development. In addition, the modest coefficient of overlap (or are close), there is a low incidence of egg
correlation can be explained by the composite nature viability on the overall developmental success. Figure 5
of survival rate that includes the capacity of the egg to (A and B) shows that a high incidence of egg viability is
survive in the water (i.e., egg viability), to be fertilized observed for a significant number of females. It should be
Fish Physiol Biochem (2021) 47:671–679 677
Fig. 4 Different causes and relative importance of egg quality fertilization. Non-malformed alevins with full completion of yolk-
defects calculated using the VisEgg and embryonic monitoring. sac resorption are considered viable alevins. Alevins with notice-
The data represent the average percentage values calculated from able morphological malformations (spinal cord torsion, head or
105 egg batches followed individually. Viable and nonviable egg caudal fin malformations, etc.) and dead alevins between hatching
data were obtained with the VisEgg system. Fertilized eggs and and resorption are considered nonviable alevins (*nonviable or
survival at eyeing, hatching, and yolk-sac resorption (YSR) are malformed alevins)
expressed as a percentage of the initial number of eggs used for
noted that the importance of egg viability on the overall 80% hatching rate; Fig. 5B). In 15.2% of the cases, there
success of development is highly variable as illustrated by is more than 10% of difference between red and blue
the significant differences between red and blue dots that dots. The percentage of batches over 10% variation be-
can be observed regardless of the overall development tween the two calculating methods of hatching rate can
success with the exception of high-quality spawns (> reach 50% of the batches for hatching rates ranging from
Fig. 5 (A) Hatching rate calculated on the number of all eggs test followed by the Dunn test post hoc analysis. Different causes
(blue) or on only viable eggs (red). (B) Percentage of batches with of egg quality defects for FISH A, B, C, and D were represented in
≥ 10% variation between the two calculating methods of hatching Fig. 6
rate. Statistical differences were evaluated by the Kruskal-Wallis
678 Fish Physiol Biochem (2021) 47:671–679
then euthanized by immersion in a MS-222 solution at a concen- Heath DD, Heath JW, Bryden CA, Johnson RM, Fox CW (2003)
tration of 400 mg L−1 (anesthetic overdose) in 3 min. Rapid evolution of egg size in captive salmon. Science. 299:
1738–1740. https://doi.org/10.1126/science.1079707
Jastrebski CJ, Morbey YE (2009) Egg size variation in lake trout:
phenotype–habitat correlations show an effect of rearing
References environment. Trans Am Fish Soc 138:1342–1351.
https://doi.org/10.1577/t08-175.1
Kobayashy W (1985) Electron microscopic observation of the
Alderdice DF (1988) Osmotic and ionic regulation in teleost eggs
breakdown of cortical vesicles in the chum salmon egg. J
and larvae. Fish Physiol 11:163–251. https://doi.org/10.1016
Fac Sci Hokkaido Univ 24:87–102
/S1546-5098(08)60200-9
Kuchnow KP, Scott JR (1977) Ultrastructure of the chorion and its
Blaxter JHS (1969) Development: eggs and larvae. Fish Physiol 3:
micropyle apparatus in the mature Fundulus heteroclitus
177–252. https://doi.org/10.1016/S1546-5098(08)60114-4
(Walbaum) ovum. J Fish Biol 10:197–201. https://doi.
Bobe J (2015) Egg quality in fish: present and future challenges. org/10.1111/j.1095-8649.1977.tb05125.x
Anim Front 66–72:66–72. https://doi.org/10.2527/af.2015- Lahnsteiner F, Weismann T, Patzner RA (1999) Physiological and
0010 biochemical parameters for egg quality determination in lake
Bobe J, Labbé C (2010) Egg and sperm quality in fish. Gen Comp trout, Salmo trutta lacustris. Fish Physiol Biochem 20:375–
Endocrinol 165:535–548. https://doi.org/10.1016/j. 388. https://doi.org/10.1023/A:1007715621550
ygcen.2009.02.011 Lubzens E, Young G, Bobe J, Cerdà J (2010) Oogenesis in
Bromage N, Jones J, Randall C, Thrush M, Davies B, Springate J, teleosts: how fish eggs are formed. Gen Comp Endocrinol
Duston J, Barker G (1992) Broodstock management, fecun- 165:367–389. https://doi.org/10.1016/j.ygcen.2009.05.022
dity, egg quality and the timing of egg production in the Migaud H, Bell G, Cabrita E, Mcandrew B, Davie A, Bobe J,
rainbow trout (Oncorhynchus mykiss). Aquaculture 100: Herráez MP, Carrillo M (2013) Gamete quality and
141–166. https://doi.org/10.1016/0044-8486(92)90355-o broodstock management in temperate fish. Rev Aquac
Brooks S, Tyler CR, Sumpter JP (1997) Egg quality in fish: what 5(Suppl. 1):S194–S223. https://doi.org/10.1111/raq.12025
makes a good egg? Rev Fish Biol Fish 7:387–416 Springate JRC, Bromage NR (1985) Effects of egg size on early
Campbell PM, Pottinger TG, Sumpter JP (1992) Stress reduces the growth and survival in rainbow trout (Salmo gairdneri
quality of gametes produced by rainbow trout. Biol Reprod Richardson). Aquaculture 47:163–172
47:1140–1150. https://doi.org/10.1095/biolreprod47.6.1140 Van Heerden E, Van Vuren JHJ, Steyn GJ (1996) Effect of
Engelmann F, Friedel T, Ladduwahetty M (1976) The native excreta, blood and vitellus contamination on fertilisation
vitellogenin of the cockroach Leucophaea maderae. Insect success of Oncorhynchus mykiss. Aquaculture 141:173–
Biochem 6:211–220. https://doi.org/10.1016/0020-1790(76 182. https://doi.org/10.1016/0044-8486(95)01228-1
)90085-8 Yamamoto T (1962) Physiology of fertilization in fish eggs. Int
Fremont L, Riazi A (1988) Biochemical analysis of vitellogenin Rev Cytol 12:361–405. https://doi.org/10.1016/S0074-7696
from rainbow trout (Salmo gairdneri): fatty acid composition (08)60545-8
of phospholipids. Reprod Nutr Dev 28:939–952. https://doi.
org/10.1051/rnd:19880607 Publisher’s note Springer Nature remains neutral with regard to
Gray BYJ (1932) The osmotic properties of the eggs of the trout jurisdictional claims in published maps and institutional
(Salmo fario). J Exp Biol 9:277–299 affiliations.