Mechanisation applied to entomological production cannot ignore insect reactivity: a case study on Bombyx mori in the context of the ‘Serinnovation’ project

Abstract

In December 2017 the Venetian Region (local Authority), financed the creation of the Operational Group (OG) ‘Serinnovation’, within the framework of the Rural Development Plan supported by the European Community. The OG aims at coordinating and spreading innovation in sericulture through mechanisation of processes and centralisation of some rearing steps, the use of waste as by-products and traceability to promote local productions. The project acts on perceived quality by increasing the added value, through production cost efficiency, and on the recovery of the waste material for further applications (circular economy). The final target was to develop a niche-process to obtain traceable ‘Made-in-Italy’ silk for the luxury market and non-textile applications. A first strategy to increase the efficiency of the process was to build an automatic leaf cutting machine to prepare the feed for the first three instars of the silkworm (Bombyx mori Linnaeus). This new machine – based on a patent – was validated through several tests and compared with the cutting system previously used. The study was completed by a bioassay of production and survival rate associated with the introduction of this innovation. The results showed that labour saving is in the order of 10% compared to a semi-manual process, the leaf quality is not affected, survival of larvae and silk production are not significantly different from the control. This methodology is proposed as a study case for other similar mechanisation processes in entomological production, as the impact of innovations on insect physiology should be carefully considered.

Introduction

The domesticated silkworm Bombyx mori (Lepidoptera: Bombycidae) is the main insect exploited for silk production. It is a monophagous insect, and in its larval stage the diet is based exclusively on mulberry leaves (Grekov et al., 2005; Resh, 2009). During the first three instars of the larval stage, leaf should be cut into small strips to increase ingestion rate since the masticatory apparatus of young larvae is more efficient in attacking leaves from the margins (Grekov et al., 2005). Furthermore, as larvae tend to climb on desiccated leaves and look for the freshest ones, administering cut leaf strips instead of the whole leaf blade decreases the risk of losing insects that tend to remain under the leaf blade in search of shelter and a moist environment. On the other hand, cutting the leaves into strips of varying dimensions to match the increasing insect size, is time consuming and labour intensive. Some attempts have been made in the past to replace manual cutting with simple tools and instruments using modified forage cutters. Apart from the human labour and the physical effort involved, these systems, have two fundamental shortcomings: (1) they are discontinuous, (2) the leaf undergoes significant compression, firstly due to the operator's load and, secondly, due to the piston pressure as it moves the leaf mass towards the cutting blade. This excessive pressure on the leaves results in a large loss of sap and thus in a rapid reduction of the nutritive value of the feed; in addition, there is a rapid fermentation of the leaf residues remaining in the machine gearboxes and in the feed if it is stored at low temperatures for the regular meal management. Furthermore, if the cut is not clear, the leaf is torn rather than cut, resulting in a rapid enzymatic browning, especially at the edges from which the small larvae begin their feeding activity. All these phenomena are rapid and can occur in the time it takes to prepare the leaves to be distributed at each meal. More modern machines were developed from the middle to the end of the last century, but these problems were not completely solved even with prototypes designed by experts in the field (Meneghini, 1975; Cappellozza et al., 1994). On the other hand, it is not easy to find scientific literature where clear descriptions of these machines are reported, mostly because they are artisanal or prototypes for research purposes (Bindroo and Verma, 2014; Chauhan and Tayal, 2017). Another problem is that these prototypes are often covered by patents for commercial exploitation, mostly written in Chinese (Tingjia, 2009; Fengjun, 2016; Peijun, 2017; Luo 2019); moreover, the development of this technology has been discontinued in Japan because first instar larvae have been increasingly reared on artificial diets (Resh, 2009) and no records of recent advances in this field have been found. The Council for Agricultural Research and Economics (CREA) through its Research Centre for Engineering and Agro-Food Processing has been involved in several projects on the management of crop biomass for animal feed, focusing on the preservation of the nutritional properties of vegetables (Assirelli and Santangelo, 2018). Therefore, by combining this expertise in mechanisation with the expertise on silkworm physiology of the Sericulture laboratory of the Research Centre for Agriculture and Environment (CREA), a new patent (Assirelli and Cappellozza, 2018) for a leaf cutting machine was developed; based on this achievement, a new prototype was built to facilitate leaf cutting and improve the general hygienic conditions through a rational design. Therefore, the main objective of the present work was to evaluate the process performance of the newly patented automatic cutting machine (aCM) in comparison with a popular semi-manual cutting machine (smCM), which was considered as a benchmark, because it can be easily assembled by farmers from common food slicers. This mechanisation attempt can be seen as a response to the requests of different stakeholders interested in a possible restoration of a silk supply chain in Italy (and possibly Europe). Mechanisation is a key point to rationalise and improve the overall efficiency of the agricultural part of the silk production chain and was a request of agricultural enterprises committed in the production of fresh silk cocoons that represent the raw material for textile and non-textile applications (e.g. cosmetics, biomedical) of silk proteins.

Materials and methods

Silkworm rearing

In the present work, two experimental theses were compared: (1) mulberry leaf strips obtained with the aCM and (2) mulberry leaves processed with the smCM. Silkworms were reared on mulberry leaves according to established best-practices (Lim et al., 1990; Grekov et al., 2005). Briefly, relevant details of the experimental setup were as follows: for each thesis, the authors reared three batches of 1000 silkworm larvae from the 1st to the 3rd instar. The silkworm larvae in both theses were then fed with whole leaf blades from the beginning of the 4th instar, when their masticatory apparatus was sufficiently developed to feed on intact leaves. For both theses 250 and 200 larvae per replicate were reared in the 4th and 5th instar, respectively. The experimental groups were reared separately until the spinning stage and cocoons from different batches were evaluated for their economic parameters. The feeding regimen of the experimental setup was three leaf distributions per day instead of the usual four, in order to increase the impact of possible drawbacks of the cutting procedure on leaf drying and thus, on development of the silkworm throughout the larval stage; to ensure a homogeneous amount of feed, the leaves fed to each batch were weighed before feeding. Three leaf distributions per day were also maintained in the last two instars, as well as leaf weighing before administration. The silkworms used in the present experiment were a commercial four-way hybrid [(121 × 125) (74 × 118)], produced by the CREA Sericulture Laboratory. The numbers in brackets indicate the progressive number of parental strains in the CREA silkworm germplasm collection.

Description of the machines

Automatic cutting machine

The aCM is based on the patent filed by CREA (Assirelli and Cappellozza, 2018) and was built by the company Connect (Borgoricco, Padua, Italy). The machine, in its first version, consisted of a main frame made of steel sections on which the various devices were mounted. The main apparatus consists of an electrical motor, a transmission belt, a cutting system, a superior feeding system and an extraction system (fig. 1). The upper part of the main frame houses two counter-rotating shafts on which three parallel cutting systems are mounted. Each cutting system has a cutting width, of 3, 6 and 9 mm, respectively, designed to produce leaf strips for feeding the silkworm in each of the first three larval instars (fig. 2). The cutting width was set according to the past experience with the smCM. The drive system consists of a single-phase electric motor with a power of 1.5 kW and a main transmission with a toothed belt that drives the two shafts on which the cutting tools are mounted. Two sets of sharp discs mounted on the shafts form a single cutting system and the contrast between the outer edges of the discs of one shaft and those of the opposite shaft produces the cut, so that the thickness of the discs precisely determines the desired cutting width (fig. 1C). There are no sharpening systems, and the cutting efficiency is guaranteed by the hardness of the material used to manufacture the discs and the precision of the mechanical assembly. A gravitational infeed system is used, in which the leaf insertion point was moved away from the cutting system for safety reasons: leaves are inserted from above into a plastic guide (infeed system) and guided by gravity through a reduced width inlet to the counter-rotating cutting system (fig. 1A). The mobile single feed system installed on the machine defines, by its positioning, the desired cutting width. To maintain the efficiency of the machine and to avoid any blockage, there is an extractor on each outer side of the cutting system (fig. 1C and 1S), which continuously cleans the cutting tools by removing the leaf strips from the grooves of the discs; the strips then fall into a plastic container placed under the cutting system. Other technical details of the cutting system are covered by the CREA patent mentioned above.

Figure 1. Overview of the automatic leaf cutting machine (aCM; A) with the feeding system visible on the right; in B the three parallel cutting systems that can be seen during end-of-season cleaning procedures; in C the details of one cutting system with its extractors (see also fig. 1S in Supplementary materials).

Figure 2. Leaf strips of different dimensions obtained from cutting systems of different width with the aCM; from top to bottom: 3, 6, and 9 mm, respectively.

Semi-manual cutting machine

This machine is a modified electric food slicer that has been routinely used by CREA. It was based on empirical experience rather than a specific project and its sole purpose was to increase productivity compared to fully manual systems while maintaining a clean cut. The slicer was adapted for leaf cutting by the addition of a plastic tube, which was used to guide the leaves towards the circular blade of the slicer using a manual wooden plunger (fig. 3). The blade of this machine must be sharpened by hand before each use.

Figure 3. Semi-manual cutting machine based on a customised food slicer. A wooden plunger is used to press leaves against the circular blade and avoid any risk for the operator.

Data collection and analysis

Physiological parameters of the silkworm

In the present work, the authors considered two important physiological parameters related to the fitness of the insect, namely the larval survival rate (LSR) and the pupae survival rate (PSR). LSR is expressed as a percentage and was calculated independently for different larval instars: from hatching to the 4th instar by counting larvae just after the end of the 3rd moult and this was defined as LSR of the 3rd instar larvae (LSR3). The same was done at the end of the 4th moult (LSR4) and at the end of the 5th and last instar (LSR5, spinning stage). PSR is expressed as a percentage and was calculated as the ratio between the live pupae inside the cocoons one week after the end of the spinning stage and the 5th instar silkworms that started to build their cocoons at the end of the 5th instar (table 1). Data on the length of the different instars were recorded as well.

Table 1. Effect of feeding silkworms with leaf strips obtained from an automatic (aCM) and a semi-manual cutting machine (smCM) on the survival of larvae and pupae, as well as on the commercial characteristics of cocoons and silk. Please refer to the text for the definitions of the acronyms

All the results are given as mean ± SE (n = 35). No significant differences were detected for any of the variables examined at the Wilcoxon–Mann–Whitney test.

Silk and cocoon economic parameters

Cocoon weight (CW), cocoon shell weight (CSW) and silk ratio (calculated as CSW/CW) were measured in order to assess the productivity of silkworms fed on both systems separately for males and females. Indeed, there are known differences in productivity between males and females that need to be taken into account for a correct assessment (Saviane et al., 2014) and thus, males were separated from females according to external morphological features on the pupal abdomen (Sakaguchi, 1978). An experimental single cocoon reeling machine was also used to determine the length of the silk thread (LST, in metres) and the titre in deniers (TST, i.e. the weight in grams of 9000 m of silk thread) (Lee, 1999).

Comparison of cutting systems

Working time analysis. Three samples of 1 kg of fresh leaves, were used per cutting dimension and per machine, giving a total of 18 samples. To calculate the working time parameters and to reduce the possible bias related to individual skills, different tests were carried out with an expert operator. The following data were recorded during the tests:

1. 1) Effective working time (EWT): expressed in seconds. It refers to the time required to process one standard unit of product, equivalent to 1 kg, starting when the operator starts to pick the leaves from their basket and ending when all the leaves have been processed.

2. 2) Accessory time (AT): expressed in seconds. It refers to the time required to prepare and clean the machine at the start and at the end of the operation, respectively.

3. 3) Total working time (TWT): the sum of EWT and AT.

4. 4) Work efficiency (WE): expressed as a percentage; it is calculated as (EWT/TWT) × 100.

5. 5) Production loss (PL): for both cutting systems, an important indicator of their performance is the leaf loss due to cutting operations. Part of the leaf waste for both machines is made up of strips that are retained in the moving parts and cannot be used due to an excessive sap loss after mechanical compression. The remaining waste consists of strips that fall to the ground and must be discarded for hygienic reasons. It was calculated as a percentage loss for each strip dimension used to feed the silkworms.

Cutting cleanliness and strip shape analysis. An image-based analysis approach was implemented to assess the uniformity and cutting cleanliness of the slices. A set of 90 leaf strips for each dimension for both systems was scanned at a resolution of about 25 px mm⁻², including a calibration scale bar in the captured images. To facilitate subsequent image processing steps, the strips were placed on a white board and illuminated with diffuse white light. The images were then converted into a greyscale and binarised into a black and white 16-bit images by setting a threshold at the inflection point of the cumulative grey scale distribution. Finally, images were appropriately resized and calibrated to remove x−y crosstalk distortions (Marinello et al., 2009). Finally, the contours of the leaf strips were extracted by implementing of a normalised gradient filter (kernel area 2 × 2 px) and converted into a set of x−y coordinates, taking advantage of the implementation of the commercial software ImageJ through the available tool ‘Analyse Line Graph’ (Schneider et al., 2012). A representation of image processing is proposed in fig. 4. All the 540 images were processed in order to estimate length and average width (L and AW respectively; see later in this section). In addition, the authors proposed a roughness analysis based on leaf strip contour to evaluate the cleanliness of the cutting of the leaf strips. Such an analysis was based on the assumption that a clear linear and regular cut is recommended to maintain the quality of the leaf strips; conversely chipping or irregular shaping of strip edges can lead to a faster deterioration of the leaves. In order to have a significant number of samples covering the typical variability of the cutting process, 96 leaf strips (16 for each machine and strip dimension combination) were randomly selected and analysed by contour roughness analysis. The contours coordinates allowed the estimation of the following indices:

• • average width (AW): the mean value of the short side width recorded at three different points (one central and two lateral positions);

• • longitudinal variation (LV): calculated, for each strip, as the ratio [Max(width)–Min(width)]/AW of the short side;

• • actual perimeter (AP) and area (AA): calculated respectively as the actual length of the extracted contour and the area contained within the same contour;

• • theoretical perimeter (TP) and area (TA): calculated respectively as the perimeter and the area of a regular rectangle having dimensions corresponding to AW and L;

• • Exceedance of perimeter (ExP), estimated as a percentage according to the following equation ((AP–TP)/TP) × 100;

• • Ratio between perimeter and area (RPA): this parameter is calculated as the ratio between the actual values of perimeter and area [AP/AA].

Figure 4. Different phases of leaf strip processing: (A) captured RGB image; (B) grey scale conversion; (C) binarisation; (D) contour extraction; (E) profile isolation; (F) roughness index estimation.

The first two parameters are self-explanatory, while the last two require some further explanation. In the case of a perfect cut, the ExP index converges to 1; as the cutting performance deteriorates, the index tends to increase, up to values that can exceed 60–80% in the worst cases. The RPA values somehow measure the availability of access points to the leaf, which is approached by the silkworms from the edges (Tsuneto et al., 2020; Shii et al., 2021). To some extent, high RPA values promote silkworm access to food, but on the other hand excessively high values could lead to faster oxidation processes and degradation of food quality.

Each strip contour allowed the extraction of two cutting cross-sectional profiles. These profiles allowed the estimation of a further root mean square (RMS) roughness index, quantified as the standard deviation of the strip edge from the mean line of the same edge.

Shelf-life assessment. Shelf-life was assessed by preserving the strips at 10 and 25°C, which, for the leaf, represent respectively the normal conditions for medium-term storage and the temperature at which larval meals occur. Five samples of leaf strips (100 g each) were evaluated per cutting dimension and machine, for each environmental condition, giving a total of 60 samples. Leaf deterioration due to water-loss was determined by weighing samples at different intervals (1, 2, 3, 4, 6, 24, 48, 72 h) after cutting.

Data analysis

For each variable examined, a pairwise comparison was performed between aCM and smCM, distinctly for each leaf strip dimension. Since many groups of measurements did not meet the normality and variance equality conditions, the Wilcoxon–Mann–Whitney test was adopted.

Results

Performance analysis

The working times of both pieces of equipment are shown in fig. 5 in the form of histograms, distinctly for the cutting of small, medium, and large strips. The total working time (TWT, top left chart) was marginally reduced with the aCM across all strip dimensions, especially with the smaller strips; however, even though a trend was clearly observable, the differences were not statistically significant in any instance, at least with the quantity of leaves that have been processed in the trials used to compare the machines. A comparable trend was noted for the AWT (top right chart), where the differences were also not significant.

Figure 5. Analysis of processing times for the automatic (aCM, grey bars) and semi-manual machine (smCM, white bars) in cutting leaf strips of varying widths (small, medium, and large). TWT, total working time; AWT, actual working time; WE, work efficiency; PL, production loss. Error bars represent standard error. For each variable and strip dimension the difference between aCM and smCM was tested using the Wilcoxon–Mann–Whitney test.

In the case of the WE (bottom left chart), the two machines exhibited no significant differences, except for cutting the wider strips, where aCM demonstrated superior efficiency, albeit with weak significance (P = 0.08).

The most significant differences between the two machines are observed in the analysis of production losses (PL, bottom right chart). The aCM exhibited a significant reduction in losses when cutting medium and large strips, with losses that were, respectively, two- and three-time larger with the smCM, with a slight significance (P < 0.1). The two machines did not differ when cutting small strips.

Geometric features of leaf strips

The two machines are designed to produce uniform leaf strips in three different sizes. Some visual examples of these strips are shown in fig. 2.

From the perspective of cutting precision and uniformity, the distribution of the AW of the strips (fig. 6) clearly shows that the automatic machine is capable of producing strips with well-distinguished sizes, tightly centred around the target width. In contrast, the semi-manual machine produces strips with much more heterogeneous widths, to the point that the target sizes sometimes overlap. This heterogeneity is particularly noticeable in the wider strips.

Figure 6. Short-side average width (AW) distribution for both cutting systems (aCM, automatic cutting machine; smCM, semi-manual cutting machine) for each cutting width (small, medium and large in red, green and blue, respectively). Data on 90 samples were recorded for each condition. The distribution highlights the greater precision of the aCM. The line represents the median value, the box the 25–75 percentiles and the whiskers the non-outlier range.

The cutting dimension analysis indicates that the two machines exhibit no significant differences in AW or AA when producing medium-width strips. However, significant differences were observed for both small and large strips (fig. 7). The small strips produced by the automatic machine are narrower and have a smaller area than those produced by the semi-automatic machine, while the opposite is true for the widest strips.

Figure 7. Comparative analysis of the geometric properties of the strips produced by trimming the leaves with the automatic machine (aCM, grey bars) or the semi-manual machine (smCM, white bars). AW, average width; LV, longitudinal variation; AA, actual area; RPA, ratio perimeter/area; ExP, exceedance of actual perimeter over theoretical perimeter. Error bars stand for standard error. Significant differences are marked with ‘*’ (P < 0.05), ‘**’ (P < 0.01), ‘***’ (P < 0.001) or ‘P < 0.1’ (Wilcoxon–Mann–Whitney test).

The aCM performs better in terms of LV, as it produces longitudinally more uniform strips, resulting in a more regular shape across all three types of cuts, ranging from 9.7 to 17.7%, compared to 19.3–26.0% for the smCM (data not shown).

The ExP graph (fig. 7) shows the variation of this indicator, which reflects how much the strip perimeter deviates from that of regular rectangles. In both machines, ExP tends to increase with the size of the strips. The automatic machine deviates more from the rectangular shape than the semi-automatic machine in small and medium strips, but in the wide strips, the deviation is lower.

RPA is also a shape indicator, specifically indicating how ‘compact’ a given shape is: for the same area, elongated shapes (i.e. with a predominant longitudinal length yielding a larger perimeter) will have a higher RPA, than shorter ones. As the strip size increases, the perimeter-to-area ratio decreases, but this happens differently between the two machines. In small strips, the aCM tends to produce more elongated strips, also as a result of the fact that the AW is smaller in this machine. In the medium and large strips, the values are more comparable, although still statistically significant.

Significant correlations between the actual perimeters and the areas of the strips were also analysed, with aCM showing a higher R ² coefficient for these correlations. This indicates a stronger relationship between area and perimeter in aCM compared to smCM, which means that the strips maintain their shape as the dimension grows.

Cutting profile cleanliness

The cleanliness of the cutting profile of the strips was assessed as a geometric property, with the relative values summarised in table 2. The RMS index is generally lower for the smCM strips compared to the aCM strips; however, the differences did not reach statistical significance for any dimension.

Table 2. Parameters about cutting profiles cleanliness for the different experimental conditions

SE, standard error.

The RMS value with its coefficient of variation (CV) describes how much the real profile of the leaf edge deviates from a straight line as described in fig. 4.

Shelf-life analysis

The strips stored at room temperature, from both the semi-manual and automatic cutting machines, obviously, exhibited faster water loss compared to those stored in the refrigerator. Strips stored at room temperature lost 10% of their initial weight approximately 6–10 h after cutting, whereas the same loss occurred in refrigerated strips after 48 h (fig. 8). After three days (72 h), the average weight loss in the refrigerator was about 15% for both machines with no differences for all strip sizes, except for the small strips, where a difference of less than 1% resulted significant at all measurement times. At room temperature, it is observed that the strips obtained with aCM show a slower water loss compared to smCM, less pronounced on the smaller strips but particularly evident on the medium and large ones. After 72 h, the losses are up to 50% for smCM and up to 45% in aCM.

Figure 8. Kinetics of weight loss due to evaporation of leaf strips prepared with two devices. (aCM, grey symbols; smCM, open symbols). Error bars stand for standard error (in most cases the bars are shorter than the symbol diameter and thus not visible). The strips were of various sizes (small, medium, large) and had been kept at two different temperature conditions (10 and 25°C). At each measurement, the differences between the two machines were statistically tested (Wilcoxon–Mann–Whitney test), and in case of a significant difference, it was indicated on the figure (‘P < 0.1’ = P < 0.1, ‘*’ = P < 0.05; ‘**’ = P < 0.01; ‘***’ = P < 0.001).

Silkworm performances

Table 1 illustrates the survival and commercial characteristics of the cocoons as a result of feeding silkworms with leaves that have been cut using two distinct machines. Commercial characteristics (mainly cocoon weight and reelability) are of paramount importance because silkworm farmers are paid according to them. No significant differences were found between the batches of silkworm larvae fed on strips from the automatic and semi-manual cutting machines. The weights of the cocoons, the percentage of silk and the length of the reeled silk thread remained unaffected by the leaf cutting method, as well as the cocoon reeling results. Data about the length of the instars are not shown since there were no differences between the experimental conditions.

Discussion

The objective of this study was to assess the performance of a new automated cutting machine for processing mulberry leaves into silkworm feed. Specifically, the study aimed to evaluate the performance of a new machine compared to a semi-manual machine that was customised from a modified food slicer. While there is existing literature on the automated control of the rearing environment, very few studies have investigated the application of mechanical technologies to leaf processing for silkworm larvae feeding. Previous patents have proposed machines for automating leaf cutting, but no studies have compared the performance of patented systems with traditional ones (Tingjia, 2009; Fengjun, 2016; Peijun, 2017; Luo 2019).

The reduction of manual labour in agriculture, particularly in livestock production, is critical for increasing process efficiency. Recently, the importance of operator safety has increased, alongside the potential for improved work organisation. Numerous works have been proposed in recent decades to mechanise various tasks, such as animal feeding, resulting in significant reductions in labour requirements. In our trials, the results demonstrated that the new aCM brings many safety and organisational advantages with improvement perspective in time saving, since the working time was generally 10% faster than the old one in processing 1 kg of fresh leaves. Under our experimental conditions, which were conducted on a small scale in a laboratory setting, the aCM did not show statistically significant improvement over the traditional equipment, which was operated by a skilled technician within a constrained timeframe. However, we believe that the difference between the machines may become more evident at the production level, when operators with less training, work longer hours in more intricate processing activities, where the ability to provide a safer and simpler task implementation may be advantageous. Additionally, the amount of product lost during raw material processing should be considered to improve overall sector efficiency (Lockheed et al., 1987). Although research on the efficiency of animal production is well-established, limited studies have focused on feed processing losses. The results of our tests indicated that the new aCM lost about half as much product compared to the smCM, thereby enhancing overall efficiency (Gaines et al., 2012; Rahmathulla and Suresh, 2012; Saviane et al., 2014; Patience et al., 2015; Schader et al., 2015; Samami et al., 2019).

Fatigue and safety hazards for farm workers (Hard et al., 2002; Kise et al., 2005; Zhao and Kong, 2010; Sahu et al., 2013; Lu et al., 2018), particularly in livestock farming (Rautiainen et al., 2004; Mitloehner and Calvo, 2008; Jakob et al., 2012) where manual tasks are still common, are important areas of research. Unfavourable working conditions have been shown to cause physical injuries that require treatment (Jakob et al., 2012). One disadvantage of the smCM that needs to be considered is the physically demanding cutting process, which strains the operator's shoulder joint. Additionally, as the feed requirement increases, the operator's performance decreases in an effort to meet the daily feed quantity for the young silkworms. In contrast, the automatic machine reduces fatigue by allowing the operator to load only a limited amount of leaves at a time and relying on the electric motor for the demanding cutting process (Kononoff et al., 2003; Schadt et al., 2012; Kornfelt et al., 2013; Giora et al., 2022). Studying feed utilisation efficiency and the effects of feed particle size and shape on animal metabolism, production performances, and survival have been previously conducted (Lim et al., 1990; Grekov et al., 2005). In terms of animal welfare indicators, such as silk production and survival rate, the new automatic machine yielded similar results to the benchmark system that has been in use in recent years. This study also investigated strip homogeneity using a novel digital image analysis procedure. The results demonstrated that the new aCM produced a more standardised product compared to the smCM. Ideally, leaf strips should have optimal dimensions for larval instars (3, 6, and 9 mm), with reduced variability. Furthermore, the operator of the smCM needs to be trained to achieve good results, while the aCM is highly accurate regardless of the operator's experience, allowing for the use of seasonal workers with little experience for a wider range of tasks. The high variability in strip width produced by the smCM confirmed this aspect, indicating the influence of individual skills, machine drives, and applied force. Additionally, the leaf strips cut with the smCM have overlapping sizes, compared to the aCM which produces strips that are easier to separate by size. The high correlation (R ²) between perimeter and area of the strips produced by the aCM indicates greater consistency. The statistically higher RPA value for the small strips cut with the aCM means that there are longer edges per unit area, providing a facilitated access to the leaf for the first instar larvae. The roughness data (table 2) demonstrated that the old cutting system has the smoothest edges with the lowest RMS values, but this parameter was independent of cutting width within the same system. Shelf-life was the final aspect considered in comparing the two machines. Aside from expected differences between different temperatures (25°C and 10°C), differences in weight loss dynamics were observed at the highest temperature, which is relevant to larval feeding conditions. However, these differences became significant only for medium and large strips within a time frame that is comparable to the time gap that occurs between two meals. This means that larvae fed with medium and large strips have fresher feed at their disposal before being fed again. For the smaller strips, differences in dehydration dynamics became relevant after 48 h.

Conclusion

After an in-depth analysis and comparison of the performance of the new aCM and the older smCM, the superiority of the former is clear due to several factors. While the strips it produces have a regular shape with more edges per unit area, which is a crucial point for the young larvae, especially in the 1st instar when they are fed with the small strips, the performance of the new cutting system is independent of individual skills which is a major advantage as it allows untrained seasonal workers to be employed, while the machine's design also meets safety requirements. The time saving and reduction in fatigue, even when processing the small quantities of leaves used in this experimental setup (compared to real requirements), are remarkable and this without affecting leaf shelf-life, larval survival rate and cocoon production. In general, mechanisation and automation of tasks in sericulture are considered as secondary issues by scientific research (Giora et al., 2022), but, as part of an action aimed at relaunching the sericulture supply chain, where the rearing of the first three instars is done cooperatively (Tassoni et al., 2024), the present innovation can be considered as a clear improvement for nurseries in charge of rearing young silkworms. At present, the Italian silk supply chain is not mature and at an embryonic stage but the aCM has already been produced and sold to a small number of farmers at the national level. They are acting as cocoon producers or egg producers and nurseries and mechanisation is a key step to achieve economic sustainability. Obviously, this approach needs to be extended to the whole agricultural process with many critical issues yet to be overcome like cocoon processing at different levels and mulberry leaves collection just to mention a few main aspects.