Background & Summary

Bite force is a performance trait which may decide on an animal’s ability to acquire food, win inter- and intra-specific fights, and successfully reproduce1,2,3,4,5. In vertebrates, maximum bite forces are well studied across a wide diversity of taxa such as bony fishes6,7, crocodilians (e.g.8), birds (e.g.9), turtles (e.g.10), squamates (e.g.11,12), frogs13, marsupials14, and mammals (e.g.15,16,17,18). Fundamental knowledge on the variation, predictors, and evolution of bite forces within the omnipresent insects is, however, lacking. Even though more than half a million insect species belong to orders that possess biting-chewing mouthparts19,20, existing literature only yields maximum bite force measurements on five dragonflies21,22, one cockroach23, 12 ants24,25,26,27, and 14 beetles28,29. This is despite the fact that biting-chewing insects include the most destructive plant-feeding animals and occupy crucial roles in the world’s ecosystems as soil-building detritivores30.

So far, measuring bite forces of insects was hampered by their small size, but the recently published measurement setup “forceX”31 overcame this limitation to some extent by allowing minimally invasive in vivo bite force measurements of animals with gape sizes more than ten times smaller than previous setups (e.g.32). Using forceX, we measured bite forces of 654 insect species from 111 families in 13 orders, collected on four continents and from numerous breeding cultures. Instead of gathering maximum force values only, as most previous bite force studies have done (but see21,22,23,29,33), we also recorded force curves. In addition, the bite force database contains head, thorax, wing, and body metrics of each specimen to assess morphological predictors for bite force in insects. Thus, the database will facilitate investigations on the macroevolution of maximum bite forces, bite lengths, bite frequencies, muscle activation patterns, and bite curve shapes across the megadiverse insects and will facilitate comparisons with all biting metazoan taxa.

Methods

Collection and material

A total of 1906 bite force series from 1290 insect specimens representing 654 species in 476 genera in 194 subfamilies, 111 families, and 13 orders were collected in Australia, China, Denmark, France, Germany, Greece, Panama, and Slovenia using light traps, insect nets, pitfall traps, or directly by hand. All specimens were collected under the respective regulations in effect (see Acknowledgements). Additionally, we measured specimens from numerous scientific, private, and commercial insect breeders and traders (Supplementary Table 1 and Acknowledgements).

Size measurements

Head width, head length, head height, thorax width, forewing length, and body length measurements were performed to the nearest 0.01 mm using a digital caliper (77001, Wentronic GmbH, Braunschweig, Germany). For the head width, the longest distance from left to right was measured, including protruding eyes if applicable (Fig. 1a). Head height in orthognathous insects was measured from the clypeo-labral suture to the dorsal end of the head (Fig. 1a,b). In prognathous insects, head length was measured from the clypeo-labral suture to the posterior end of the head (Fig. 1c). Thorax width was measured on the prothorax (Fig. 1d) and excluded lateral protrusions as found e.g. in many cockroaches and praying mantises. Body length measurements excluded cerci, ovipositors, or other abdominal appendages (Fig. 1e).

Fig. 1
figure 1

Insect length measurements. (a,b) Head of an orthognathous insect in frontal (a) and lateral view (b). (c) Head of a prognathous insect in lateral view. (d) Frontal part of an orthognathous insect in dorsal view. (e) Habitus of an orthognathous insect in lateral view. Abbreviations: bl: body length; hh: head height; hl: head length; hw: head width; tw: thorax width; wl: forewing length. a,b,c after Snodgrass50; d after own photo, e after Snodgrass51. Figures not to scale.

Full size image

Bite force measurements

All measurements were carried out with the metal-turned version of the forceX setup as described in31. In short, live and conscious animals were held between two fingers, rotated by 90° along their body axis and allowed to voluntarily bite on the tip elements of the forceX. Different tip element designs31 and distances between them were used to accommodate different animal gape sizes. During measurements, animals were observed through the Junior Stereo 3D microscope (Bresser GmbH, Rhede, Germany) that is part of the forceX setup to ensure that gape sizes are suitable and that the insects bite at the edge of the tip elements so that the ratio of the forceX lever remains at a constant 0.53831,34. We also checked if the animals bit with the distal-most incisivi of their mandibles to ensure that measurements remain comparable31,34. Non-distal bites or wrongly placed bites on the tip elements were discarded. If animals did not start biting by themselves, we used the tip element protrusions to insert the tip elements between the mandibles and/or used a fine brush to touch the animal’s cerci, head or abdomen23. Amplified analogue voltage signals were converted to a digital signal by a 12-bit USB data acquisition device (U3-HV, LabJack Corporation, Lakewood, Colorado, US) and recorded with the LJStreamUD v1.19 measurement software (LabJack Corporation) on a computer.

Data curation

Subsequent data curation was performed in the software environment ‘R’ v. 4.2.235 using the package ‘forceR’ v.1.0.2031. Since the forceR package was written to analyse data generated with the forceX setup, we used, if not stated otherwise, the default settings of the package functions. After initial download of the raw data from Zenodo via zen4R v.0.836, time series were converted from the output format of LJStreamUD to a *.csv file containing only a time and a voltage column (without changing measurement values) using the forceR function ‘convert_measurement()’. Then, all measurements were manually cropped using ‘crop_measurement()’ to exclude regions without bite data at the beginning and end of each measurement. Next, ‘amp_drift_corr()’ was used to correct for the logarithmic drift of the analogue charge amplifier (see31 for details). When using the high amplification setting (20 V/N) to amplify the miniscule voltage signals of the piezoelectric force transducer at small bite forces, the zero-voltage-line (‘baseline’) may drift notably during a measurement. Therefore, a PDF file depicting all input raw data and their amplifier drift-corrected data graphs (available at Zenodo, s. Data Records) was visually inspected, and, if necessary, the function ‘baseline_corr()’ was used in its automatic mode to correct for this drift. In some of these cases, however, especially when the test animals showed long, plateau-like bite curve shapes, the automatic mode of ‘baseline_corr()’ failed to find the baseline, and the manual mode was used. All corrections can be retraced in the PDF file and reproduced using the log files that were created during corrections and which are stored at Zenodo. With the function ‘reduce_frq()’ we then reduced the sampling rate of all time series to 200 Hz, a value found to be sufficient to represent insect bite force curves21,22,23,29 to reduce the amount of data for further analyses. As a last curating step, voltage values were converted into force data [N] with the forceR function ‘y_to_force()’ that considers the amplification level of each measurement and the lever mechanics of the measurement system. Supplementary Table 1 shows all measurement settings, taxonomic classifications and information about which correction procedures have been performed on which measurements.

Maximum force value extraction for specimens and species

To extract maximum force values of each specimen and each species and calculate the standard deviations of these values we used the function ‘summarize_measurements()’ of forceR and custom code, relying on the ‘dplyr’ v.1.1 package37. We then plotted the log10-transformed average maximum bite force per specimen (grey dots in Fig. 2) and per species (black dots) against the log10-transformed average body length (Fig. 2a) and head width (Fig. 2b) using ‘ggplot2’ v.3.4.238 and ‘ggExtra’ v.0.939. Linear regressions through the log10-transformed species-wise data showed a significant positive relationships between body size and head width (p < 0.001) with explanatory values of R² = 0.44 and R² = 0.56 respectively. Due to the expected logarithmic releationship between size and bite force40, means were calculated as geometric means. Calculations with the regular mean, however, yielded similar results (p < 0.001, R² = 0.44 and R² = 0.56; Supplementary Fig. 1). Following existing literature, we use the term “maximum bite force” for the force that was produced by the animals during our experiments.

Fig. 2
figure 2

Maximum bite force against body length (a) and head width (b). Grey dots show values of all measured specimens, black dots show geometric means per species. Marginal histograms at the x- and y-axes demonstrate mean size and mean bite force distribution per specimen, respectively. Regression lines and coefficients refer to log10-linear models of species-wise bite force against body length (a) or head width (b). Orange diamonds in (b) show bite force measurements available in previous literature. All axes are log10-transformed.

Full size image

Comparison to previous insect bite force measurements

Previous studies on insect bite forces covered maximum bite force values for 32 species21,22,23,24,25,26,27,28,41. To check if these measurements follow similar allometric slopes as our data, we extracted all available insect bite force data from the literature and added them to the scatterplot in Fig. 2b, excluding data on 12 ant species because head width was not reported25,26,27 or force was produced by mandible kinematics, not muscles41. We then tested if our data and the literature data differ in their allometric slopes by comparing a linear model with the null hypothesis of different slopes (log10(bite.force)∼ log10(head.width) * source) versus a linear model with the null hypothesis of common slopes (log10(bite.force) ∼ log10(head.width) + source). Both model fits were compared with an ANOVA to find out if they differ significantly.

Assessment of geographical coverage

Climate zone data (Köppen–Geiger classification system42,43,44) was gathered for each species based on the GPS coordinates of its collection localities (Supplementary Table 1) with the function ‘LookupCZ()’ of the R package ‘kgc’ v.1.0.0.245. Percentages of species in the database for each country and climate zone were calculated.

Assessment of phylogenetic coverage

To assess the phylogenetic coverage of the bite force database we compared the number of species with database entries to the number of species listed by the Open Tree of Life46, accessed on 2022/02/05 with the function ‘tol_node_info()’ of the package ‘rotl’ v.3.0.1447. Comparisons were carried out for all insect orders and families that are present in the bite force database.

Data Records

All raw measurements and the cleaned data form the database at Zenodo48. All data files are available in comma-separated format as single files and in a combined long format converted to a 200 Hz sampling rate. Additionally, the database contains PDF plots and log files created during the conversion of the raw data to the final database. Supplementary Table 1 is also stored in the same repository.

Technical Validation

Visual inspection of the scatter plot of bite force against head width (black dots in Fig. 2b) and all literature data points (orange diamonds) revealed that the literature data lies close to the regression through all data points of our database. This impression is corroborated by the comparison of the allometric slopes of the insect bite force database and the literature data, which yielded no statistically significant difference (ANOVA: F = 0.107, p = 0.74).

Geographical assessment of the collected animals showed that most species of the insect bite force database were collected in Australia (31.0%), Germany (19.1%), and Panama (16.2%). 23.2% of the species were obtained from breeding cultures. The remaining 10.4% of the species were collected in Greece, Slovenia, France, China, and Denmark. Climate region assessment revealed that most species were collected in temperate (53.7%) and tropical (43.5%) regions. 2.8% came from dry and continental regions combined (Fig. 3b). We did not consider the original geographic distribution of those species obtained from breeding cultures.

Fig. 3
figure 3

Geographical and phylogenetic coverage of the bite force database. (a) Species entries per collection location (country or breeding). (b) Species entries per Köppen–Geiger climate zones42,43,44 with specimens sourced from breeding cultures excluded. (c) Ratio of database entries compared to species estimated in all insect orders present in the database. (d) Ratio of database entries compared to species estimated in all insect families present in the database. The dashed lines in (c,d) mark a ratio of 1 data base entry per 100 estimated species. Abbreviations: Am: tropical monsoon; Aw: tropical savanna with dry-winter characteristics; As: tropical savanna with dry-summer characteristics; BSh: semi-arid (steppe) hot; Csa: mediterranean hot summer; Csb: mediterranean warm/cool summer; Cfa: humid subtropical; Cfb: oceanic; Cwa: dry-winter humid subtropical; Dwb: warm summer continental.

Full size image

A total of 13 biting-chewing insect orders are present in the database (Fig. 3d). We could not obtain live animals from the orders Zoraptera and Grylloblattodea. Bite force measurements of the few species of Plecoptera, Mecoptera, and Trichoptera that were available failed because no voluntary biting could be elicited in these specimens. We did not attempt measuring available representatives of Psocoptera and the biting-chewing “mandibulate archaic moths” (Lepidoptera: Micropterigoidea) due to their minute size. The assessment of phylogenetic coverage of the orders and families showed that most families are represented by less than one species entry per 100 estimated species (Fig. 3d,46). While orders were sampled in proportion to their taxonomic diversity (Fig. 3c), we were only able to measure at least 1% of the described species in Mantophasmatodea, Phasmatodea and Mantodea (dots left of dashed line in Fig. 3c). Accordingly, bite forces of only a fraction of all insect species were measured so far. Nevertheless, the database exceeds all previous studies combined in species numbers (20-fold in insects, 3.5-fold in amniotes), marking just the beginning of research on this performance trait in the most species-rich metazoan clade.

Usage Notes

The forceR package31 was used to create the insect bite force database, which contains cleaned measurement time series and maximum bite forces of insects. The same package may be used to expand the scarce knowledge on insect bite forces by tackling questions regarding the evolution of bite lengths, frequencies, and bite curve shapes by semi-automatically extracting individual bite curves from these measurements. Additionally, the maximum bite force values presented in Supplementary Table 1 can be used for a wide range of in-depth studies on the morphological and ecological predictors and macroevolution of this important performance trait in the megadiverse insects.