Exploring Bruchus rufimanus Egg Substage Dynamics at Different Constant Temperatures

ABSTRACT

The broad bean weevil Bruchus rufimanus is a major pest of faba bean in Europe, where infestations are responsible for increasing economic losses. Despite its importance for both predictive modelling and biological control strategies based on oophagous parasitoids, the thermal biology of egg development remains poorly understood. This study aimed to address this gap by collecting pods bearing eggs and incubating them under 11 constant temperatures ranging from 12°C to 40°C. Egg development through four morphological substages was monitored daily. Data were organised as stage‐frequency matrices and life‐tables' parameters were estimated through an optimisation problem that reconstructed the distributions of stage‐specific development times. Early substages (S1–S2) were generally longer at lower temperatures, whereas differences among substages became minimal within the optimal thermal range of 20°C–35°C. At 40°C, development ceased at stage S3, preventing egg hatching. The proposed method provided reliable estimates of development time distributions and showed consistent thermal responses across substages. These findings provide the first quantitative baseline for the egg development of B. rufimanus and offer valuable parameters for physiologically based models. Moreover, they supply essential information for synchronising parasitoid releases with susceptible egg stages, thereby supporting the development of sustainable integrated pest management strategies.

1. Introduction

The broad bean weevil Bruchus rufimanus Boheman, 1833 (Coleoptera: Chrysomelidae), is a major insect pest of faba bean ( Vicia faba L.) crops and is responsible for increasing economic losses across Europe (Simmen 2020; Segers 2023). As with many insect pests, the life cycle of this univoltine species is strongly affected by environmental conditions and host plant phenology (Hamidi et al. 2021; Segers, Noël, et al. 2023). Under typical conditions, gravid females lay their eggs on developing green pods at the beginning of the growing season, allowing larvae to subsequently develop within the maturing seeds (Howe and Currie 1964; Kingsolver 2004). The preimaginal development of this pest is cryptic and seed damage generally becomes evident only upon adult emergence from infested beans. Over the last decade, B. rufimanus has attracted increasing attention in Europe because of the marked rise in infestations and the lack of effective control strategies currently available (Ward 2018; Segers 2023).

The development of B. rufimanus causes damage to faba bean seeds at multiple levels. Larval feeding activity reduces seed germination rates, leads to the accumulation of waste material, which affects the organoleptic properties of the seeds and increases the risk of fungal contamination during storage (Kaniuczak 2004; Khelfane‐Goucem and Medjdoub‐Bensaad 2016). In addition, infested seeds show reduced aesthetic quality, often resulting in the rejection of batches from food and feed markets.

Although recent studies have investigated the field bio‐ecology, infestation dynamics, chemical ecology and varietal resistance of B. rufimanus , no study has yet explored egg development under laboratory conditions (Segers et al. 2022; Segers, Carpentier, et al. 2023; Segers, Noël, et al. 2023). The main reason for this gap in knowledge lies in the difficulty of establishing continuous laboratory rearing of this species under controlled conditions due to the obligatory reproductive diapause that occurs during the overwintering period. Under natural conditions, B. rufimanus overcomes this mandatory diapause of 6–7 months through a specific combination of photoperiod (16 light hours) and a nectar‐ and pollen‐based diet (Tran and Huignard 1992; Tran et al. 1993).

Segers et al. (2024) proposed a new methodology for rearing this pest under laboratory conditions and, for the first time, described the total egg incubation period (from oviposition to egg hatch) across 11 constant temperatures ranging from 12°C to 40°C. In addition, they showed, with supporting images, that the egg development is characterised by four macroscopically distinguishable morphological stages:

1. freshly laid egg with a homogenous whitish opaque appearance (S1),

2. the mid‐grey stage, corresponding to the formation of the first larval tissues (S2),

3. the black‐head stage, corresponding to the cephalic capsule of the larvae (S3) and

4. the hatched egg characterised by the typical exit hole on the pod surface (S4).

The first three substages correspond to the embryonic development, while the fourth stage (the hatched eggs) corresponds to the beginning of the first larval instar.

The morphological distinction between the egg substages can provide valuable insights into the timing of egg hatching over temperature, thereby complementing the information already reported by Segers et al. (2024). Accordingly, this study aims to investigate and analyse the thermal biology of the egg substages, providing a quantitative baseline for the parametrisation of mathematical models and for further exploration of egg–parasitoid interactions. In addition, this work represents an example of a life‐table study in which the data were collected as stage‐frequency matrices, since technical constraints prevented the tracking of individual egg histories. Eggs are laid in clusters and strongly attached in large numbers to the pods, making individual monitoring particularly challenging. Furthermore, their very small size makes repeated observations difficult within reasonable time constraints. These technical limitations led to the development of an innovative estimation method (Rossini, Cotorruelo, et al. 2026) capable of reconstructing the distributions of the stage‐specific development times from the information contained within the stage‐frequency matrices.

2. Materials and Methods

2.1. Eggs Collection and Incubation Under Different Constant Temperatures

The data collection process of this study was based on the rearing protocol described by Segers et al. (2024). Batches of pods bearing freshly laid eggs were placed in climatic chambers (Snijders/Fisher scientific S.R.L.) to evaluate their thermal response at 11 constant temperatures: 12.4°C ± 0.4°C, 14.9°C ± 0.9°C, 17.7°C ± 0.4°C, 20.5°C ± 0.5°C, 22°C ± 1°C, 23°C ± 2°C, 23.4°C ± 0.6°C, 28.0°C ± 0.6°C, 31.1°C ± 0.4°C, 35.0°C ± 0.4°C and 40.0°C ± 0.4°C. Each egg‐bearing pod was inspected daily to track egg development across four morphological stages. Host plants did not receive any treatments that might interfere with the results, as detailed in Segers et al. (2024).

The operational difficulties behind the rearing (i.e., the impossibility of isolating single eggs from the pod) did not allow for tracking the life history of each individual, according to the standard protocol described in Rossini et al. (2024) and Rossini, Lots, et al. (2025). Accordingly, data were collected and organised as stage‐frequency matrices for development and mortality (presented in Data S1), also defined as integral representation of life‐tables dataset (Rossini, Lots, et al. 2025). This representation requires the data analysis protocol described by Rossini, Sweidan, et al. (2026) to reach the stage‐specific parameters typical of life‐table studies. The key steps of the data analysis applied are detailed in the section below.

2.2. Data Transformation and Analysis

The development periods of the different egg substages were obtained by an optimisation problem which estimated the distribution of the stage‐development times from the stage‐frequency matrices (Rossini, Cotorruelo, et al. 2026). Reminding the most interested readers of the definitions in Rossini et al. (2024), Rossini, Bono Rosselló, et al. (2025), Rossini, Lots, et al. (2025), Rossini, Cotorruelo, et al. (2026), Rossini, Sweidan, et al. (2026), we hereby report only the backbone of the analysis, fully included in the Matlab (vers. 2024b, Mathworks) script publicly available at https://github.com/lucaros1190/BRufimanusEggSubstages. Two steps were foreseen in this phase of the study: (i) obtaining the population dynamics of the four egg substages with respect to the oviposition time (absolute time) and (ii) applying the optimisation problem to estimate the stage‐distributions of the development times.

2.2.1. Step 1: From Stage‐Frequency Matrices to Input/Output Signals

Stage‐frequency matrices describe the number of individuals $I_i\left(t\right)$ and $D_i\left(t\right)$ that, at each sampling time $t$, are alive or dead, respectively, in the egg substages $i=S1,S2,S3,S4$. The single substage can be read as a compartment (Manetsch 1976; Rossini et al. 2019, 2020) featured by an incoming flux of individuals $y_{i-1}\left(t\right)$ coming from the previous life stage and two outcoming flux of individuals $y_i\left(t\right)$ and $d_i\left(t\right),$ which denotes the individuals that reach the next life stage and that die, respectively. The variation of individuals alive $I_i\left(t\right)$ and dead $D_i\left(t\right)$ into the stage $i$ between the time $t-1$ and $t$ is defined as follows:

$$\mathrm{\Delta }{I}_i\left(t\right)=I_i\left(t\right)-I_i\left(t-1\right)$$ (1a)

$$\mathrm{\Delta }{D}_i\left(t\right)=D_i\left(t\right)-D_i\left(t-1\right)$$ (1b)

The outcoming flux of individuals that develop to the next stage over time is given by:

$$y_i\left(t\right)=y_{i-1}\left(t\right)-\mathrm{\Delta }{I}_i\left(t\right)-d_i\left(t\right)$$ (2)

where

$$d_i\left(t\right)=\mathrm{\Delta }{D}_i\left(t\right)$$ (3)

The Equations (2) and (3) conclude the first step of the data analysis by providing the population dynamics of the four egg substages referred to the oviposition time.

2.2.2. Step 2: From Input/Output Signals to the Distributions of the Stage‐Development Times

The second step, instead, estimates the distributions of the development times from the input/output signals (the population dynamics) calculated with Equations (2) and (3), to further compute the life‐table statistics. The differential representation of life tables corresponds to the knowledge of the impulse response $w_i\left(t\right)$ (Rossini et al. 2024; Rossini, Lots, et al. 2025) which connects the input and the output signals computed through the Equations (2) and (3). Accordingly, the distributions of the development times can be estimated via the following Quadratic Programming (QP) problem:

$$\begin{array}{c}\begin{array}{c}\underset{\begin{array}{c}{w}_i\left(0\right),\dots ,w_i\left(\infty \right)\hfill \\ w_{d,i}\left(0\right),\dots ,w_{d,i}\left(\infty \right)\end{array}}{\min }\sum _{t=0}^{\infty }\alpha {‖{\widehat{y}}_i\left(t\right)-y_i\left(t\right)‖}^2+\beta {‖{\widehat{d}}_i\left(t\right)-d_i\left(t\right)‖}^2\\ \text{Subject to}:\\ \begin{array}{c}{\widehat{y}}_i\left(t\right)=\sum _{\tau =-\infty }^t{\widehat{w}}_i\left(t-\tau \right)y_{i-1}\left(\tau \right)t=0,\dots ,\infty \hfill \\ \begin{array}{cc}{\widehat{d}}_i\left(t\right)=\sum _{\tau =-\infty }^t{\widehat{w}}_{d,i}\left(t-\tau \right)y_{i-1}\left(\tau \right)& t=0,\dots ,\infty \end{array}\\ \begin{array}{c}\begin{array}{cc}{w}_i\left(t\right)\ge 0& t=0,\dots ,\infty \end{array}\\ \begin{array}{cc}{w}_{d,i}\left(t\right)\ge 0& t=0,\dots ,\infty \end{array}\\ \begin{array}{cc}\sum _{t=0}^{\infty }{\widehat{w}}_i\left(t\right)+{\widehat{w}}_{d,i}\left(t\right)=1& t=0,\dots ,\infty \end{array}\end{array}\end{array}\end{array}\hfill \end{array}$$ (4)

The optimisation problem (4) also accounts for the typical features of the distribution of the stage‐development times, namely the positivity of the distributions and the mass conservation between the number of dead and alive individuals. The mean, standard deviation, median, mode, kurtosis and skewness associated with the estimated distributions of the stage‐development times have been computed according to the standard procedures defined in Rossini, Lots, et al. (2025).

3. Results and Discussion

The estimated distributions of the development times at the different constant temperatures are presented in Figures 1, 2, 3, 4 for the stages S1–S4, respectively. Differences in distribution shape were observed both among the egg substages and across temperatures, reflecting distinct developmental patterns and thermal responses. These findings are further supported by the summary values reported in Table 1, which presents the main life‐table parameters computed for the differential representation.

FIGURE 1.

Estimated differential representation of the Bruchus rufimanus embryonic development substage S1.

FIGURE 2.

Estimated differential representation of the Bruchus rufimanus embryonic development substage S2.

FIGURE 3.

Estimated differential representation of the Bruchus rufimanus embryonic development substage S3.

FIGURE 4.

Estimated differential representation of the Bruchus rufimanus embryonic development substage S4.

TABLE 1.

Common life‐tables parameters computed for the differential representation, including mean, median and modal development time (days), kurtosis and skewness.

Temperature (°C)	Egg substage	Mean ($\pm$SE)	Median	Mode	Kurtosis	Skewness
12.4 ± 0.4	S1	10 $\pm$ 2	11	11	−0.248	−1.046
	S2	17 $\pm$ 2	15	15	−0.055	0.766
	S3	4 $\pm$ 1	5	5	1.035	−1.742
	S4	8 $\pm$ 3	8	6	0.784	0.505
14.9 ± 0.9	S1	7 $\pm$ 1	8	8	−1.529	−0.214
	S2	13.3 $\pm$ 0.7	13	13	2.26	2.064
	S3	7.9 $\pm$ 0.5	8	8	16.851	−4.341
	S4	3 $\pm$ 2	3	3	0.043	0.782
17.7 ± 0.4	S1	6.4 $\pm$ 0.8	6	6	0.656	0.976
	S2	6 $\pm$ 1	6	6	5.969 108	1.850 104
	S3	2.00 $\pm$ 0.01	2	2	7.003 105	812.637
	S4	3.0 $\pm$ 0.3	3	3	883.015	22.444
20.5 ± 0.5	S1	4 $\pm$ 0	5	4	—	—
	S2	6 $\pm$ 1	3	7	−1.490	−0.0672
	S3	1.0 $\pm$ 0.3	2	1	115.096	10.821
	S4	2.4 $\pm$ 0.6	2	2	21.493	1.954
22 ± 1	S1	5.4 $\pm$ 0.5	5	5	−1.959	0.201
	S2	3.1 $\pm$ 0.3	3	3	7.268	2.346
	S3	2.3 $\pm$ 0.5	2	2	−1.430	0.754
	S4	1.7 $\pm$ 0.5	2	2	−0.502	−0.053
23 ± 2	S1	4 $\pm$ 1	5	5	−1.993	−0.078
	S2	3 $\pm$ 1	3	3	−1.296	0.202
	S3	2.00 $\pm$ 0.02	2	2	−1.320	0.824
	S4	6.0 $\pm$ 0.4	—	—	−1.011	−0.206
23.4 ± 0.6	S1	3 $\pm$ 0	3	3	—	—
	S2	4.8 $\pm$ 0.4	5	5	0.875	−1.398
	S3	1.000 $\pm$ 0.002	1	1	9.310 108	2.916 104
	S4	2.1 $\pm$ 0.4	2	2	11.480	2.333
28.0 ± 0.6	S1	4 $\pm$ 0	4	4	—	—
	S2	1.8 $\pm$ 0.3	2	2	2.450	−1.936
	S3	1 $\pm$ 3	1	1	2.165 108	1.414 104
	S4	2.00 $\pm$ 0.03	2	2	1.235 104	107.088
31.1 ± 0.4	S1	2 $\pm$ 0	2	2	—	—
	S2	2.00 $\pm$ 0.01	2	2	2.958 104	146.159
	S3	2.00 $\pm$ 0.02	2	2	1.544 104	116.919
	S4	1.00 $\pm$ 0.02	1	1	1.420 104	116.031
35.0 ± 0.4	S1	1 $\pm$ 0	1	1	—	—
	S2	2.6 $\pm$ 0.5	3	3	−1.825	−0.381
	S3	1.000 $\pm$ 0.002	1	1	3.959 109	5.300 104
	S4	2.1 $\pm$ 0.3	2	2	8.671	3.266
40.0 ± 0.4	S1	1 $\pm$ 0	1	1	—	—
	S2	4.99 $\pm$ 0.02	5	5	1.545 104	−117.336
	S3	1.00 $\pm$ 0.03	1	1	1.426 104	116.041
	S4	—	—	—	—	—

Proceeding by order, the substages S1 and S2 were generally longer than the subsequent substages at lower temperatures. Differences in development time among substages became much less pronounced within the optimal thermal range (25°C–40°C; Segers et al. 2024), although substage S2 consistently remained slightly longer than the others. In contrast, the development times of substages S3 and S4 showed lower variability across temperatures overall, with only a slight increase at lower temperatures compared with optimal and higher thermal conditions. It is worth noting that eggs exposed to 40°C were unable to hatch, as development did not progress beyond substage S3.

Median and mode values reflect the same trends observed in the mean development times, suggesting that the estimated distributions were generally regular. Kurtosis and skewness, instead, showed a more variable pattern, likely reflecting some limitations of the estimation algorithm described in Equation (4). From a theoretical perspective, the algorithm assumes that the differential representation of life tables corresponds to the stage‐impulse response, thereby neglecting the developmental history of individual organisms. Although this approximation may be biologically plausible, it should be further investigated and validated using larger datasets, potentially allowing the introduction of correction terms where necessary.

Life‐tables studies are fundamental for understanding the responses of terrestrial arthropods to environmental conditions. In the case of insect pests, this information represents a key component of pest management programmes, as it supports the estimation of the parameters required by physiologically based models. Notably, development, mortality and fertility rates are commonly derived from life‐tables data (Rossini, Bono Rosselló, et al. 2025; Rossini, Sweidan, et al. 2026; Baser et al. 2025; Rossini and Bruzzone 2025; Corbetta et al. 2026). Although such studies have been conducted for many insect species, relatively little is known about the embryonic response to temperature (Kutcherov 2020). Nevertheless, this information can be crucial for the development of multitrophic models (i.e., host–parasitoid systems), as it provides quantitative insights into the processes occurring between oviposition and egg hatching.

Understanding insect egg development is crucial for characterising interactions between egg masses and oophagous parasitoids, which represent valuable biological control agents capable of preventing infestations at their earliest stage. In the case of bruchids, whose larvae develop endophytically, oophagous parasitoids are particularly promising because they can prevent egg hatching and subsequent infestation of seeds. However, parasitisation rates may vary considerably depending on environmental conditions, egg substage and the emission of volatile organic compounds involved in parasitoid attraction (Sabbatini‐Peverieri et al. 2020).

A detailed understanding of egg development can therefore improve biological control strategies by supporting a better synchronisation of parasitoid releases with the most susceptible embryonic stages, as well as by improving mass‐rearing protocols prior to augmentative releases in the fields (Bittau et al. 2021). To date, only two egg–parasitoid species have been identified for B. rufimanus : Uscana semifumipennis Girault, 1911 and U. senex (Grese, 1923), both belonging to the family Trichogrammatidae (Chalcidoidea). Nevertheless, little is currently known about their ecology, host preferences or foraging behaviour. Research has mainly focused on egg parasitoids of another storage pest species named Callosobruchus maculatus (Van Alebeek 1996), whereas most studies on other bruchids have concentrated on larval rather than egg parasitoids. Useful insights may also be drawn from other open field pest–parasitoids systems, such as Trissolcus japonicus (Ashmead, 1904), which parasitise Halyomorpha halys (Stål, 1855) eggs at early developmental stages, typically no older than 1–3 days (Qiu et al. 2007; Sion et al. 2025). In this context, the present study provides baseline information on the duration of the different egg substages as a function of temperature, offering valuable indications for identifying the interaction windows that may maximise parasitoid efficiency.

In conclusion, this study provides important quantitative information on the developmental timing of each embryonic stage of B. rufimanus in relation to temperature. These biological data represent an essential basis for future investigations into the interactions between this pest and its egg parasitoids, with the broader aim of identifying new biological control tools within an integrated pest management framework.

Author Contributions

Conceptualisation: Arnaud Segers and Luca Rossini. Data curation: Arnaud Segers and Luca Rossini. Formal analysis: Luca Rossini and Arnaud Segers. Funding acquisition: Frederic Francis, Rudy Caparros Megido and Arnaud Segers. Investigation: Luca Rossini and Arnaud Segers. Methodology: Arnaud Segers, Luca Rossini, Rudy Caparros Megido, Frederic Francis and Emanuele Garone. Project administration: Arnaud Segers, Rodrigue Lugendo, Rudy Caparros Megido and Frederic Francis. Software: Luca Rossini and Arnaud Segers. Supervision: Rudy Caparros Megido, Emanuele Garone and Frederic Francis. Validation: Rudy Caparros Megido, Emanuele Garone and Frederic Francis. Visualisation: Arnaud Segers and Luca Rossini. Roles/writing – original draft: Arnaud Segers, Luca Rossini. Writing – review and editing: Arnaud Segers, Rodrigue Lugendo, Luca Rossini, Rudy Caparros Megido, Grégoire Noël, Emanuele Garone and Frederic Francis.

Funding

This research was funded by Service Public de Wallonie Agriculture, grant number D31‐1395″. L.R. is funded by the European Commission under the Marie Sklodowska Curie Actions Postdoctoral Fellowship (MSCA‐PF‐2022) project ‘PestFinder’, Grant no. 101102281.

Disclosure

All authors agree to the submission of the present manuscript, which has not been published before and is not concurrently being considered for publication. No generative AI was used at any stage of this study, including writing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: ppl70986‐sup‐0001‐Supinfo.xlsx.

Data Availability Statement

The scripts and the datasets to fully reproduce the results of this study are publicly available at https://github.com/lucaros1190/BRufimanusEggSubstages.