Larval material was obtained either by direct breeding of wild collected adults by the authors or donated by other scarab breeders for the purpose of this study: 2 last instar larvae of G. albosignatus Boheman, 1857 donated by O. Jahn (Czech Republic), having been reared from beetles imported from Tanzania in 2004; 12 last instar larvae of G. goliatus (Drury, 1770) reared from adults imported from Cameroon in December 2010; 6 last instar larvae of G. orientalis Moser, 1909 donated by O. Jahn (Czech Republic), having been reared from beetles imported from Tanzania in 2004.

The terminology for larval description follows Hayes (1929), Böving (1936) and Ritcher (1966). Antennomeres I–IV were abbreviated in the description with ‘an I’ – ‘an IV’. In order to provide the most accurate information on chaetotaxy, the hair-like setae of the cranium and other structures, were classified by their relative size into two groups: medium to long (80–300 µm) and minute or small (5–40 µm or less). Morphological observations and measurements were made using a Nikon SMZ 745 stereomicroscope and Olympus BX 40 dissecting microscopes, equipped with an Olympus Camedia 5060 digital camera. Photographs were taken using a Canon 70D digital camera, equipped with a Canon MP-E 65/2.8 macro lens with 5:1 optical magnification and a Canon EFs 60/2.8 macro lens for images of larger body parts. Partially focused images of each specimen were combined using Zerene photo stacker software (Zerene systems LLC, Richland, USA). All pictures were digitally enhanced using Adobe Photoshop CS4.

The specimens included in this study are deposited in the following collections:

CUPC Department of Zoology, Charles University, Prague, Czech Republic (Petr Šípek)

NMPC National Museum, Prague, Czech Republic (Martin Fikáček, Jiří Hájek)

For these experiments, larvae obtained by breeding two pairs of goliath beetles (G. goliatus) imported from Cameroon in January 2009 were used. They were kept together in a breeding terrarium (90 × 45 × 55 cm) with a 30 cm deep mixture of soil and leaf litter. The substrate was checked once a week and the newly laid eggs were transferred individually to 500 ml plastic boxes for hatching. The larvae were kept in the same boxes during the entire first and second instar. Third instar larvae were transferred to 1000 ml plastic boxes. Larvae were raised in separate containers during the entire experiment to prevent cannibalism and to allow individual tracking of growth. The breeding substrate was composed of a mixture (1:1) of crushed beech (Fagus sylvatica) leaf litter and organic soil (common garden compost). Approximately half of the substrate was replaced with fresh substrate every weighing period. Boxes were kept in a climate chamber at an average temperature of 28°C with a 12:12 L/D cycle. Water was added to the substrate when necessary and the substrate was kept damp but not sopping. The eggs were monitored every other day to determine the date of hatching and newly hatched larvae were randomly divided among three diet regimes.

To examine the dependence of larval development on nutriment, larvae of G. goliatus were reared under three different dietary regimes: 1) on substrate with proteins added ad libitum (‘fully nourished regime’, 23 larvae); 2) reared on substrate, but proteins were supplied after a period of starvation (‘partly nourished’, 11 larvae); and 3) reared on substrate without the addition of proteins during the entire experiment (‘undernourished regime’, 11 larvae); see below. Some of the larvae were killed at the end of the experiment and used for the study of intestinal microorganisms (Zadrobílková et al. 2016).

The rearing conditions of the initial two instars were identical for all larvae in the experiment. In accordance with the breeding manual (McMoningle 2001, 2012; Meier 2003; Meier and Campbell (undated)) we started to feed all larvae on pellets of soft-moist dog food (FROLIC® Complete with Beef) from the onset of the second instar. These pellets were replaced every weighing session to prevent an excessive growth of mites and other unwanted organisms. In order to monitor the effect of nourishment on larval growth, one cohort of larvae was allowed to continue to feed on pellets (‘fully nourished group’) from the onset of the third instar whereas the other larvae were denied pellets. A part of the unfed larvae were allowed to resume feeding on pellets at a given point after a period of protein deprivation during the third instar (100–240 days after the onset of the third instar). These individuals are referred to as the ‘partly nourished’ larvae. The last cohort of larvae (‘undernourished group’) was raised without a supply of pellets during their entire final instar.

To monitor larval development, we weighed larvae every five days from hatching throughout their entire development using a KERN 450-3M digital scale with a precision to 0.001 g. This weighing interval was chosen in view of the optimal frequency of pellet replacement (McMoningle 2006, 2012; Meier 2003, Meier and Campbell (undated)), which minimizes larval stress and the proliferation of scavenger mites and moulds. For each instar, we determined development time (in days) and maximal mass (in milligrams). In order to compare growth under different feeding regimes in the third instar, we calculated the growth rate at the beginning of the instar and the growth rate just before and after the start of pellet supply to the partly nourished larvae. Growth rate was calculated as a daily mass increment measured over a period of ten days in all aforementioned periods. The initial weight, first recorded at the beginning of the instar was not taken into account because the larvae moult with an empty gut which is refilled after ecdysis, thus growth rate computed with this initial weight would contain false growth caused by new gut content. The recording period before pellet supply was ten days immediately prior to protein addition, after pellet supply it was from the fifth to the fifteenth day after protein addition (pellet supply). The calculating period of growth rate for undernourished larvae was arbitrarily set as the 150^th – 160^th day after the beginning of the instar which corresponded roughly to the point when the feeding of the partly nourished larvae with proteins began.

To compare the development times and body mass under the food regimes of the first two instars and the final instar we used a one-way ANOVA and Student’s t-test, respectively. As initial weight is expected to be correlated with growth rate, differences in growth rate were tested using an ANCOVA with the initial weight of the recording period as covariate. Normality of the data was verified using the Kolmogorov-Smirnov test, the Cochran test indicated that variances were homogeneous so no transformations were necessary. The significance level was set to 0.05. Statistical analyses were performed using the program STATISTICA, version 6.0 (StatSoft Inc. 2001).

Breeding conditions during the first and second instar were identical for all larvae; therefore development times and maximal larval mass of these larval stages are presented as a whole irrespective of the experimental regime (Table 3). Uniformity of the development times and maximal body mass in the first two instars under all three experimental regimes was also confirmed by a one-way ANOVA (p > 0.4 for development times and body mass in both instars). Total development time of the fully nourished larvae was 193 ± 5 days. The larvae spent the longest period of time in the third instar, which lasted roughly twice as long as the previous two instars combined. Also, most of the growth took place in the final instar, as the larvae gained 80% of their final body mass during this time (Table 3).

Food manipulation had a considerable effect on survival and growth. None of the eleven starved larvae pupated, whilst 20 out of the 23 larvae (87%) reared under the fully nourished regime and four out of the eleven larvae (36%) reared under the partly nourished regime pupated; this difference was statistically significant (two-tailed Fisher’s exact test: p < 0.01). On the other hand, all larvae died during the prepupal stage in the pupal cell.

In the third instar, there were clear differences in growth trajectories between the breeding regimes (Fig. 4). Whilst fully nourished larvae grew regularly till construction of the pupal cell, larvae reared under the other two regimes grew only for a short period after the onset of the third instar (although considerably slower than fully nourished larvae; F_{1, 42} = 99.1, p < 0.001; see also Table 4). The larvae subsequently stopped growing and remained around the reached weight (9969 ± 435 mg; the values were not significantly different between the partly nourished and undernourished regimes: t₂₀ = 1.07, p = 0.42). Nevertheless, after the addition of pellets to their diet, the partly nourished larvae were able to resume growth, although their final body mass (t₃₂ = 5.45, p < 0.001; see also Table 3) and growth rate (F_1,31 = 20.9, p < 0.001; see also Table 4) were significantly lower than those of the fully nourished larvae.

Figure 4. 

Individual growth trajectories of the fully nourished larva (red line), partly nourished larvae (black and blue lines) and undernourished larva (green line). Evidently, the absence of proteins in larval diet had profound consequences on development. In the third instar, the starved larvae were able to resume growth immediately after the addition of protein to their diet. The inset image shows mean growth of all eleven partly nourished larvae 40 days before and after pellet supply (SPS), irrespective of actual time of pellet supply. Means ± standard errors are depicted.

In the character matrix of 38 larval features published by Micó et al. (2008) no distinct character distinguishing Goliathus larvae from the other 41 cetoniinae taxa was identified (except for the evidently wrongly coded character 34/0 – abdominal segments IX and X not fused). Among the other characters shared with only a few other Goliathini or Cetoniinae larvae were the number of respiratory holes on the thoracic spiracle (state 30/2 – more than 60; shared with Mecynorhina polyphemus (Fabricius, 1781) and Fornasinius fornasinii) and the pretarsi with a sharply pointed claw bearing 2 basal setae (character state 32/0, shared with members of Valgini, Trichiini, Schizorhinini, Cremastocheilini, Fornasinius fornasinii and Dicronocephalus wallichi Hope, 1831). Similar looking pretarsi have also been described for Hypselogenia geotrupina (Oberholzer 1959), Agestrata orichalca (Linnaeus, 1769) (Zhang 1984) and members of Taenioderini (Vendl et al. 2014). While the number of respiratory holes may be a size-dependent character, the shape of the pretarsus deserves a more detailed investigation. A falcate claw with two setae at the base clearly represents an ancestral/plesiomorphic stage of pretarsal morphology in the entire plant feeding lineage of scarab beetles (Ritcher 1966, Sawada 1991) but its morphology varies considerably. There are species/groups with a highly developed claw (e.g., the apical part distal to the setae is longer than the proximal part, e.g., the Trichiini), or there are species with an extremely reduced apical portion of the claw (most Taenioderini) and all possible transitional stages occur. McMonigle (2006) argued that the claws of the genus Goliathus are retractable and compared them with the claws of Mecynorhina (Mecynorhina) torquata (Drury, 1782) and Mecynorhina (Megalorrhina) harrisii Westwood, 1847. He stated that only the (sub)genus Megalorrhina possess claws similar to Goliathus, although less developed. We were not able to confirm that the claws of Goliathus are “retractable”, but it is clear that the claws are capable to fold back against the tibiotarsus, which seems unusual among pleurostict scarabs. In general we may state that the development of the pretarsal claw in Goliathus is exceptional, in G. goliatus and G. orientalis the claw is even equal in length to the tibiotarsus.

Other distinct characters of Goliathus larvae include the extraordinarily coarse surface of the cranium and the extremely well developed stemmata (larval eyes). The most striking feature is the general habitus of living larvae which are straight (Fig. 1C), unbent and have a relatively slim appearance (e.g., the posterior part of the abdomen is more or less as thick as the other parts of the abdomen and thorax) compared with the size of the imago. The characteristic C-shaped form, which is common to most other scarab larvae (living, non-moving), is not present here (unless killed and preserved using standard methods (Figs 1A, B)). Several other larvae of Coryphocerina (Cetoniinae: Goliathini) also have a similar slim appearance, but tend to be C-shaped when in a resting position in substrate (personal observation). The extremely well-developed mandibular scissorial teeth of Goliathus larvae (the first tooth is falcate) are also rather extraordinary and exceptional among Cetoniinae (although with this character more caution needs to be exercised as these teeth are often abraded in older larvae). Similarly, the sharply pointed external tooth of the mandible may be another unique character, although there are species with a similar but more or less blunt tubercle: Hypselogenia geotrupina, Ichneostoma pringlei Perissinotto et al., 1999, Rhomborhina polita Waterhouse, 1875 (Oberholzer 1959, Sawada 1991, Perissinotto et al. 1999) and others.

Several species-specific characters have been identified in the immature stages of G. albosignatus, G. goliatus, and G. orientalis, most of them distinguishing G. albosignatus from the other two species (see Table 2), which may be due to the different size range of this species. Nevertheless a few characters distinguishing G. goliatus from G. orientalis have been identified which is quite surprising as these species are closely related and even hybridizing to the F1 generation in captivity (McMonigle 2006, Meier and Campbell (undated)). The validity of these characters still needs to be confirmed, but they might be regarded as additional support for the current separation of the species-level classification of G. goliatus and G. orientalis.

Although there are no data available on larval biology and development of goliath beetles in the wild, thanks to the long-standing efforts of beetle breeders some interesting findings about their developmental requirements in captivity are available. One of these is the presumed obligatory requirement of proteins in larval diet during its development (McMonigle 2001, 2006; Meier 2003; Meier and Campbell (undated)). Our results confirm this statement. The growth rate of larvae without added protein clearly slowed down immediately after the exclusion of protein pellets at the beginning of the third instar and from a certain point they were not able to achieve a higher body weight (the average weight after ecdysis was around 6 g, while the threshold weight when the larvae ceased their growth was around 10 g). This was in contrast to larvae which were fully nourished with proteins, most of which were able to construct a pupal cell. After the addition of protein pellets to their diet the larvae clearly responded by resuming growth and, interestingly, some of them even constructed a pupal cell, some of them after up to 200 days of starvation. However, all larvae in our experiment died in the pupal cell, possibly due to a high rearing temperature. It is not clear how long the larvae could live without a protein diet, but some of them were still alive 250 days after the beginning of the third instar, when the experiment was terminated and the larvae were inspected for intestinal microorganisms (see Materials and methods). Some of the breeding manuals state that from a certain point in time the larva does not consume the substrate and feeds purely on protein pellets (Wong 2008). Nevertheless, in our experiment the starved larvae produced an amount of faecal pellets comparable to the fully nourished larvae. It is possible that in adverse conditions, when the larva cannot find a suitable source of nutrients, it feeds solely on substrate and waits for more favourable conditions when prey becomes available.

It has been suggested that goliath beetle larvae are carnivorous and prey on the larvae of other rose chafers in the wild (McMonigle 2001, 2006, 2012; Meier 2003). Indeed, in captivity goliath beetle larvae readily consume the larvae of other common species (e.g., Pachnoda; Klátil and Vrána 2008, personal observation). It is also known that other rose chafer larvae (e.g., Eudicella, Cheirolasia, etc.) enhance their diet by feeding on rose chafer larvae of other species or are even cannibalistic (Klátil and Vrána 2008, Micó et al. 2008, personal observation), but this behaviour is only facultative and the larvae are able to finish their development normally without protein input (Christiansen 2013, personal observation). The possible dependence on live prey may also be reflected in their larval morphology. Mandibles with sharp scissorial teeth (Figs 2, 3), an additional pointed tooth on the lateral face of mandible, legs with conspicuously long and pointed claws (Fig. 3J–K), and well developed stemmata may possibly be linked to a predatory way of life. It is also possible that thanks to a protein rich diet goliath beetles develop considerably faster than comparably sized scarab beetles such as Megasoma or Dynastes, which thrive well on a “classical” substrate. The development of these species normally takes up to two or three years (Glaser 1976, Morón and Deloya 2001, Klátil and Vrána 2008). However, it can be considerably faster if fed with protein-rich pellets (McMonigle pers. comm.). In this experiment, the regularly fed larvae developed in less than 200 days on average, although their maximal weight was considerably lower than the commonly reported weight (up to 100 g in large males and 50 g in females). The alleged association of the closely related Argyrophegges larvae and hyrax (Mammalia: Procaviidae) may indicate that other representatives of the subtribe Goliathina also have alternative larval feeding strategies which may be helpful in achieving their exceptionally large size.