Indigenous communities play significant role in reporting traditional uses of flora and can be used as a tool to conserve and maintain biodiversity and could be further match with scientific findings (Mishra,  2013;  Aziz  et  al., 2018). It plays an important role in the documentation of diverse plant species that are utilised by local communities for treating a broad spectrum of disorders (Shah   et   al.,   2016)   and  also  for  the  discovery  and development of novel, safe and affordable medicines (Süntar, 2020). Also the preservation, protection and promotion of the indigenous knowledge is encouraged as it plays a critical role in health and food security among other areas (UNCTAD, 2004).

Trees do not only serve a purpose of being host of edible insects but through indigenous knowledge they are used in several ways such as medicines for treating ailments, for apiculture among others and therefore, it is prudent to encourage communities to conserve these plants in order to protect habitats for edible insects like caterpillar species (Kusia et al., 2021). Considering the threat of climate change (Kalantary, 2010) and poor harvesting methods of trees (Chungu et al., 2007), the need to protect the forests is greater now than before.

Insects are the dominant class of organisms in terms of species number on a global level and almost half of them feed on living plants (Schoonhoven et al., 2005). The chemical constitution of a plant is the prime factor in its interaction with the insect world (Schoonhoven et al., 2005). The nutrition of plants may have an impact on the growth, survival and general performance of the insects that feed on plants (Shah, 2017). The main groups of primary plant metabolites; proteins, carbohydrates, and lipids involved in fundamental plant physiological processes form essential nutrients for herbivores. Therefore, qualitative and quantitative variation in primary plant compounds can have profound effects on insect preference and performance (Berenbaum, 1955).

The choice of host plants is determined both at the egg-laying and larval-feeding stages (Nishida, 2014). The food quality of a given host plant tissue will influence the performance and preference behaviour of herbivorous animals (Tremmel and Müller, 2013). Insects lay most eggs on host plants that provide the best larval performance (Griese et al., 2020).  In an event that the parent lay eggs on an unsuitable plant, larvae have the capacity to find a suitable food source (Zhou et al., 2020). This is why female herbivores try to lay eggs where their offspring can develop successfully (Li and Liu, 2015).

Plant palatability which is the preference an animal has for a particular feed when offered a choice (Heath et al., 1985; Kochare et al., 2018) plays a major role in host selection which in turn is affected by different plant factors such as seasonal availability of plant, degree of maturity, growth stage, phenology, morphological and chemical nature (Muhammad et al., 2014; Kochare et al., 2018). Food quality is important for all caterpillar instars (Arriens et al., 2021) and the community composition of insect herbivores is largely shaped by host–plant phylogeny and functional traits (Abe et al., 2021). Caterpillar species and the host plant on which they feed, have an important effect on the host acceptance and performance (Traine et al., 2021).

Plant volatiles are an important component of the wide- ranging   classes   of   plant   compounds   that   influence herbivory as many insects detect their host plants through volatile chemical signaling (Carroll et al., 2008, Magalhaes et al., 2012; Massad et al., 2017). Therefore, chemical composition of plants determines the relationship between phytophagous insects and their host plants (Kant et al., 2015). For instance, in the case of phytochemical compounds, they serve several purposes such as to repel or attract phytophagous insects (Kant et al., 2015). Phytophagous insects recognise the host plant through multiple sensory modalities, including visual, olfactory, gustatory, and tactile cues (Nishida, 2014). Herbivorous insects are known to possess mechanisms which enable them to detect and reject plants that contain harmful secondary compounds as ingestion of these compounds can result in death (Hennessy, 2020). Host plant odour is an important cue in recognising hosts suitable for oviposition (Cao et al., 2014).

Currently, there is limited knowledge on plant chemicals that influence Gynanisa maja forage preference. The aim of this study was to test whether indigenous knowledge was in accordance with the results of laboratory tests regarding the preference of the commonly consumed caterpillar G. maja (Lepidoptera: Saturniidae). In order to achieve this, we first conducted a survey to determine people’s observation on the most preferred host plants for G. maja then secondly, phytochemical and proximate contents in the host plants were analysed.

Study site

The study was conducted in Luampa district, Western province of Zambia (15° 3’ S and 24° 24’ E, 1119 m asl) in five clusters namely: Katunda, Mwandansengo, Mbanyutu, Luampa Mission and Nkenga as shown in Figure 1. According to the information obtained at Luampa District Health Office (unpublished data), Luampa has a tropical savannah climate with a hot-wet season (November to April), cold dry season (May to July) and a hot dry season (August to October). The average annual rainfall is 1000 mm. The average temperature ranges from 10 to 38°C. The main vegetation type found in this district is mixed forest commonly referred to as miombo woodland which is characterized by the dominant tree species of the genera Brachystegia, Julbernardia and Isoberlinia (Gumbo et al., 2018). This vegetation type is an important habitat for lepidopterans and other insects (Ghaly, 2010) as forest cover prove to be influential for insect communities (González et al., 2017). Soils in this area are dominated by acidic arenosols and podzols type with sandy textures and pH less than 4.0 (JAICAF, 2008).

Sampling and data collection

Household survey

To determine people’s observation on the most preferred host plants  for  G.  maja,  household  survey  was   conducted   using  a questionnaire. The survey focused on description of caterpillar morphological features, social-economic value, harvesting sites and the preferred host plants. The targeted sample include 80 randomly sampled participants per cluster who had lived in the area for 15 years or more, because these people might have harvested G. maja for a couple of years and could provide valid and sufficient information about the preferred food plants of G. maja in the area.

Tree sampling and leaf sample collection

Tree species mentioned during household survey were recorded in each cluster that is, Katunda, Mwandansengo, Mbanyutu, Luampa Mission and Nkenga. Trees (n=3) that were commonly mentioned in all the clusters were further studied in the field by collecting leaves for laboratory analysis. With the help of a key informant, one tree species per cluster was randomly sampled from farms where respondents harvest G. maja and about one thousand to two thousand moderately young leaves per tree species per cluster were collected. The leaves were randomly selected based on the information obtained from the respondents who indicated that G. maja larvae prefer moderately young leaves to old ones as they harvest more of it from moderate young leaves than old. The leaf samples  were   plucked   from   their  respective    tree   branches, packaged in well labeled polyethene bags. Thereafter, they were air dried under the shade and stored in zip lock bags for proximate and phytochemical analyses following the method by Banu and Cathrine (2015) and Sithara  et al. (2016).

Proximate and phytochemical analysis

To determine proximate chemical analysis, three host plants were analysed and the following parameters were assessed: moisture content, total ash, crude protein, fat, carbohydrate and calorific value following the methods recommended in the literature (Egan et al., 1981; AOAC, 1990; Zaklouta et al., 2011). Moisture content was determined by keeping the samples in a moisture oven at 105°C for overnight and then cooling in desiccator. The loss in weight was regarded as a measure of moisture. To determine ash content, samples were heated in a muffle furnace at 600°C, until white or greyish-white ash was obtained, and the weight of the ash was recorded directly. Nitrogen content was estimated using micro Kjeldahl method and crude protein was calculated by multiplying the measured nitrogen by a factor 6.25 (Zaklouta et al., 2011). The calorific value was calculated by multiplying the values of total carbohydrate, fat and protein by the factors 4, 9 and 4 respectively and     the    summation    was    expressed    as    kilocalories.   For phytochemical analysis, the following contents were qualitatively assessed using ethanol and methanol as solvents: the tannins, saponins, flavonoids, phenols, terpenoids, and alkaloids following the method by Jayaramu et al. (2016) with minor modifications. Salkowski reaction was used to determine the presence of steroids (Kumar et al., 2010). Therefore, general reactions in these analyses revealed either the presence or absence of these compounds in different leaf extracts.

Statistical analysis

The survey data on people’s observations on the preferred host plant species of G. maja were related to the results of quantitative proximate and qualitative phytochemical laboratory analyses. Shapiro- Wilk test was conducted on proximate data to test normality and in order to achieve normality and stabilize homogeneity of variances between treatments, the moisture data was square root transformed while ash, protein, fat and energy data were log transformed before performing one-way analysis of variance (ANOVA). The ANOVA was applied to determine whether differences in moisture content, total ash, crude protein, fat, carbohydrate, or calorific value were due to tree species and/or clusters). In order to determine where exactly the significant difference occurs between groups, Tukey's honestly significant difference (HSD) test was used. Generally proximate data were reported as mean ± standard deviation. Data were analysed in R version 4.1.1 software.

People’s observations of tree hosts for G. maja

Five host plants were mentioned as preferred hosts for G. maja, namely; Julbernadia paniculata (Benth.) Troupin, Brachystegia longifolia Benth, Erythrophleum africanum (Welw. ex Benth.) Harms and Albizia antunesiana Harms. Out of these, J. paniculata, E. africanum and B. longifolia were mentioned consistently in all the five clusters in this study. Across all clusters, 100% of the participants mentioned J. paniculata as the first host choice for G. maja, while 57.5% of the participants in Katunda, 51.3% in Luampa and 38.8% in Mbanyutu mentioned B. longifolia as the second preferred host for the caterpillar but E. africanum was mentioned as the second most preferred host only in Mwandansengo (63.8%) and in Nkenga (56.3%). On the other hand, only 17.5 and 26.3% of the participants in Katunda mentioned that B. spiciformis and A. antunesiana respectively as hosts for G. maja. Overall, three host plants were mentioned in all the clusters namely; J. paniculata (100%), E. africanum (50.3%) and B. longifolia (49.8%) (Figure 2).

Proximate composition of the preferred host tree species for G. maja

Tables 1 and 2 shows proximate composition of the preferred   host   plants  for  G.  maja  and  ANOVA  table respectively. The proximate composition of three host tree species for J. paniculata, E. africanum and B. longifolia commonly found in all clusters are presented in Table 1. The highest content of moisture (11.62 ± 0.02%) was found in E. africanum and the least amount (2.40 ± 0.0 7%) was found in J. paniculata. Total ash also varied between tree species with the highest content (4.33 ± 0.00 7%) observed in J. paniculata, and the least content (1.0 ± 0.12%) observed in E. africanum (Table 1). The differences in moisture content and total ash observed between tree species was significant (Moisture content: F_2.15 = 494.8, p < 0.001, Total ash: F_2.15 = 557.0, p < 0.001, Table 2).

Crude protein content was significantly different among the top three plant species (F_2.15 = 77.6, p < 0.001, Table 2). J. paniculata had the highest crude protein content (12.44 ± 0.07%) and the least content (0.31 ± 0.19%) was found in E. africanum (Table 1). The observation that J. paniculata scored highest crude protein content was consistent across all clusters. The highest fat content (13.54 ± 0.21%) was found in E. africanum and the least content (1.07 ± 0.55%) was detected in J. paniculata across clusters (Table 1), and the differences across species was highly significant (F_2.15= 77.9, p < 0.0001, Table 2). Furthermore, the highest content of carbohydrate (84.63 ± 0.77%) was found in B. longifolia and the least content (66.94 ± 0.29%) was detected in E. africanum. Although the least content of carbohydrate was found in E. africanum, this tree species significantly scored the highest energy content (427.66 ± 1.22 Kca/100 g) while B. longifolia was found to have the least energy content (366.78 ± 2.48 Kca/100 g) (F_2.15 = 187.9, p < 0.001, Table 2).

In fact, carbohydrate content negatively correlated with fat content (r = -0.61) and energy (r = - 0.59) suggesting that carbohydrates decreased with fats and energy across tree species studied (Table 3). But fat content was highly correlated (r = 0.90) with energy (Table 3) suggesting that fats were more important than carbohydrates for deducing the energy content in host tree species. Furthermore, protein content negatively correlated with moisture (r = -0.59) and carbohydrates (r = -0.74). But the strengths of the relation between protein and fat (r = 0.03), protein and energy (r = 0.24), protein and ash (r = 0.29) were weak.

Phytochemical analysis

Results for the qualitative analysis of phytochemicals of J. paniculata, E. africanum and B. longifolia leaf samples are presented in Table 4. Alkaloids and flavonoids were detected in all samples, suggesting that all three host tree species for G. maja contain alkaloids and flavonoids. Saponins and phenols were detected in B. longifolia and E.  africanum  but  not  in  J.  paniculata. The presence of both saponins and phenols was strongest in E. africanum than in any other tree species tested (Table 4). Saponins were not the only phytochemicals that were absent in J. paniculata, phenols and steroids were also absent. But steroids were only detected in B. longifolia and E. africanum. Tannins were however, not detected in all three host tree species. Furthermore, different extraction methods produced different results. The extraction method based on ethanol did not detect saponins and phenols compared to the methanol-based method (Table 4), suggesting that methanol-based extraction method is a better method for detecting these phytochemicals than ethanol-based method.

The proximate  composition  and  phytochemical analysis of host trees for the edible caterpillar, G. maja, in this study matches with local people’s observation on the host preference of this caterpillar. This is so as the local people mentioned that when G. maja translocate to any other host tree species from J. paniculata, the caterpillar become stunted, changes in colour from greenish to brownish and also the taste is not as good as that of G. maja that completes its entire life cycle on J. paniculata (Chanda et al., 2020).

The relatively high level of total ash detected in J. paniculata, indicates that it is rich in minerals (Ogbe and Affiku, 2011) but the quantitative profiling of minerals in the preferred hosts is still not clear. The high protein and moderately low carbohydrate content were detected in J. paniculata. This is not surprising considering that tree species with high carbohydrate and fat contents are generally low in protein (Lykke and Padonou, 2019). The growth  of  insect larvae and the length of the larval stage are critically dependent on food quality particularly those that are rich in proteins, in fact, juvenile insects require relatively large amounts of protein in their diets (Speight et al., 2008). Low fat content in J. paniculata relative to E. africanum and B. longifolia also provide a potential reason why it is most commonly observed as a preferred host. Eigenbrode and Espelie (2019)highlighted that fats in plants play an important role as epicuticular lipids on plant surfaces. The physical structure of plant surface lipids may affect insect herbivore attachment, locomotion and deter oviposition and feeding.

The presence of phytochemicals (alkaloids and flavonoids) in the host trees in this study might also explain why the three species (J. paniculata, E. africanum and B. longifolia) seem to be preferred by the caterpillar. In fact, several studies have shown that although alkaloids play a role in chemical defence against insects, they also serve as precursors for biosynthesis of male courtship pheromone components (Ackery and Vane-Wright, 1984; Schulz, 1998; Lawson et al., 2021) which attracts females in most Lepidoptera as oviposition cue (Macel, 2011). Although, alkaloids are possibly used as a general cue to recognize host trees amongst other tree species, selection depends on other factors such as nutrients  (Van  der  Meijden  et  al.,  1989;  Macel, 2011). The presence of alkaloids and superior nutritional value of J. paniculata probably make this tree the most preferred among the trees studied. Similarly, flavonoids were observed in J. paniculata and these phytochemicals are crucial for pollen development and provide a visual and volatile cues to attract pollinators and seed dispersers (Dar et al., 2017).

Terpenoids were only observed in J. paniculata and its been reported that most monoterpenoids are volatile compounds, found mainly as components of essential oils  (Schoonhoven et al., 2005),  and some terpenoids act as chemo-attractants to herbivores (Boncan et al., 2020). Considering that terpenoids also function prominently in host location by herbivorous insects (Miller and Strickler, 1984; Visser, 1986; Niogret et al., 2011). It is well known that other phytochemicals such as saponins are not desirable to insects and trees with high content of saponins and tannins are more likely to be avoided as hosts by several insects including Lepidoptera (Adel et al., 2000; Singh and Kaur, 2018; Qasim et al., 2020). When ingested, saponins increase mortality levels by lowering food intake and affecting movement of food in the insect gut due to toxicity and lower digestibility (Adel et al., 2000; Singh and Kaur, 2018; Qasim et al., 2020). Among  all  host  trees  tested,  J.  paniculata  was found with little or no saponins, this might be another reason why this host is preferred by G. maja. Furthermore, phenols were absent in J. paniculata but present in B. longifolia  and  E.  africanum. This  group  of phytochemicals in a tree act as protective agent and inhibitor against invading herbivores (Lattanzio et al., 2006; Dar et al., 2017). “These findings can provide a potential  explanation  for the fact that 100% respondents in this study mentioned that G. maja larvae prioritise J. paniculata as a more preferred host than E. africanum and B. longifolia . This provides an opportunity to preserve this tree species so as to prevent G. maja from going into extinction which may have serious consequences on household food security particularly for the local people.

The  findings  on  indigenous  knowledge  on  host plants matched well with plant chemicals wherein chemical analyses provided a potential explanation why J. paniculata was observed as a primary host of the wild edible G. maja as it contained high amounts of total ash (rich in mineral content) and crude proteins but low-fat content. These are major nutrients that influence the selection of a host as they play an important role in the growth, development, and survival of herbivorous insects. In addition to nutrients, the phytochemical presence of alkaloids, flavonoids and terpenoids as well as the absence of tannins, saponins,  phenols   and   in   J.   paniculata,  also supports the respondent’s observation as these phytochemicals play significant role of repelling or attracting herbivorous insects such as G. maja. However, future studies should focus on profiling phytochemicals beyond detecting whether they are present or absent in order to enhance the chemical ecology of G. maja. It is evident that J. paniculata is the most preferred host tree for this caterpillar, this provides an opportunity to preserve this tree species so as to avoid sending G. maja into extinct which may have serious consequences on household food security particularly for the local people.

The authors have not declared any conflict of interests.

This study was funded by World Bank through the Africa Center of Excellence in Sustainable Use of Insects as Food and Feeds (INSEFOODS) at Jaramogi Oginga Odinga University of Science and Technology. The authors appreciate the research group of the Copperbelt University in Zambia for the technical support. They are grateful too to the Ministry of Health (MoH) through the Department of Food and Drugs Control Laboratory for offering their laboratories for analysis of plant chemicals and provision of technical support. They further extend their gratitude to Eng. Nicholas Chintu Sande for providing technical support on the development of the map and Mr. Davies Muyengu, a resident of Mbanyutu cluster for his selfless support during the household survey.