2.1 Data Collection and Construction of the Database

To obtain regression parameters for insects from different orders and geographic regions, we searched the Web of Science, Scopus and Google Scholar for relevant literature in July 2024. We used the following search phrases: “length–mass AND relationship AND insect”, “length AND dry AND mass AND regression”, “body size AND biomass AND aquatic insects”, and “linear AND body AND dimensions AND mass AND aquatic AND insect”. The obtained literature was scanned for relevance, and for relevant publications, we subsequently used cross-referencing to obtain more publications.

All literature was checked for reliability by selecting peer-reviewed publications only, followed by checking for the presence of data that satisfied the following three criteria: (1) regression parameters of body length to dry mass (i.e., parameters for regressions of body lengths to wet mass or ash free dry mass (AFDM) were not included; e.g., Rosati et al. 2012); (2) regression parameters were only included for body lengths, in which body length was defined as the linear length from the front edge of the head to the tip of the abdomen, without antennae, cerci, legs, etc. Other linear body dimensions were not included, as body length is the most frequently measured variable (Burgherr and Meyer 1997); (3) regressions other than the power regression (such as linear, quadratic or exponential regression models) were not included in the database, but for readers interested in these values, we refer to the list of original publications in Table 1. After applying these selection criteria, we collated parameter values from all 25 publications that contained this type of information directly, or in their supplementary information, in the case of the reviews by Benke et al. (1999) and Johnston and Cunjak (1999). Although for some regions and specific taxa there may be more information available—for instance, in non-English or in literature not digitally available—we believe that we captured the majority of the most used information currently available in the peer-reviewed international literature.

From these 25 studies, we extracted for each taxon the following key data: the continent on which data had been collected, the country or local region of study, and the parameters for the scaling factor and exponent of the length–mass regressions. Linearised regressions were converted to power regressions to create a standardised database across all taxa, but all original parameter values are also included (to avoid under- or over-estimations due to transformations, Mährlein et al. 2016). Data were obtained from eight insect orders with aquatic life stages (Coleoptera, Ephemeroptera, Diptera, Hemiptera, Megaloptera/Neuroptera (in figures further referred to as Megaloptera), Odonata, Plecoptera and Trichoptera). We separated the data based on geographic regions representing five continents (“Asia”, “Europe”, “North America” excluding lower Mexico, “Oceania” including Australia, and “South America” including Central America), which combines an anthropogenic and geographic viewpoint on the performed research effort in different parts of the world. A full overview of extracted variables with metadata is provided in Table 2.

Regression parameters were typically derived from several specimens measured to the nearest 0.1 mm. The database includes 523 regressions, that is, each based on between n = 3 and n = 1423 specimens (median number of specimens is n = 31, mean is n = 63 ± 124 SD, with only 34 of the regressions based on less than 10 specimens). To enable readers to assess the robustness of the regressions, the number of measured specimens is provided in the database. All reported relationships were always significant at the alpha = 0.05 level, and R² values were judged as representative by the original authors. After the lengths of specimens were measured, they were typically dried at temperatures between 40°C and 100°C for minimally 24 h and then cooled in a desiccator. Subsequently, they were weighed individually with accuracies varying between 0.1 and 10 μg (Burgherr and Meyer 1997; Benke et al. 1999). We did not assess the effects of preservation methods on the regression parameters, as this was not our primary focus, but it is important to emphasise that chemical preservation methods can affect insect lengths (Benke et al. 1999; Mährlein et al. 2016). Such preservation effects may differ among taxa, types of chemicals used, duration of storage, and the size of the specimens—and for more in-depth reading, we refer to Johnston and Cunjak (1999).

To assess the current geographic and taxonomic coverage of the body length–mass regression parameter dataset and identify knowledge gaps, we compiled a comprehensive list that includes the currently known families for each relevant order and, within families, the number of known genera per continent. Since taxonomy, as well as the number of species, genera, and even families, is continuously changing, this list should be considered as the best possible estimate reflecting current taxonomic richness. We were unable to accurately determine the number of genera for families in the order Diptera, as most families are regarded as “aquatic” and encompass partly aquatic and fully terrestrial species too, and this variation can also occur within genera. Genus-level analyses for Diptera are therefore excluded. The compilation of the list of taxa was based on taxonomic expertise supported by up-to-date catalogues, other literature sources, as well as reliable online databases (as Table S1 or on Figshare, Van Leeuwen et al. 2025).

2.2 Data Analyses

Data were not statistically analysed because data availability for most family and/or geographic regions was too low to perform meaningful analyses. Instead, we visualised relative data availability and relations in figures and provided the database itself directly for readers to assess and further study possible variation for their specific taxa of interest. The parameter values for Haliplidae in Shahbaz-Gahroee et al. (2021) were not included in the database because the values were identified as strong outliers.

4.1 Variation in Data Availability

Parameter data were completely lacking for Africa and Antarctica. From the remaining continents, Oceania and South America had the fewest available parameter values at the family level. The poor representation of genus-level parameter data across our entire dataset—and even less on the species level—explains the frequent use of less precise order or family-level parameters in the literature. This variation may be due to the presence of extensive morphological variation within orders among families, or due to variation within families among genera. Order-level parameter values can provide relevant body mass estimates (Smock 1980; Meyer 1989; Poepperl 1998; Bried and Ervin 2007), but it is important to note that regression parameters can add considerable variation to data. Still, any inaccuracy of regression parameters should be seen in the light of other variations, such as inaccuracy of density estimates that may arise from variation in sampling effort (Méthot et al. 2012).

Information on genus-level parameter availability can help guide the focus of future studies. For example, for the aquatic insects described for Midwest North America (Bouchard et al. 2004), data were available for all 7 Odonata, all 8 Plecoptera and both Megaloptera families, 15 of the 17 Trichoptera, and 11 of 13 Ephemeroptera families. However, data at the family level were only available for 3 of the 13 Hemiptera families, 5 of 8 Coleoptera, and 8 of 19 Diptera families. In contrast, for insects in Northern Europe (Nilsson 1997), no European data were available for the 11 Hemiptera families. Only 2 of the 16 Coleoptera, 1 of the 3 Megaloptera/Neuroptera, and 9 of the 24 Diptera families were represented. Data availability was much better for the Plecoptera (6 of 7), Odonata (5 of 9), Trichoptera (13 of 19) and Ephemeroptera (6 of 9). This type of information can be used a priori during study designs or the selection of target species.

4.2 Variation in Regression Parameter Values

In the dataset, we detected considerable variation in parameter values, which likely reflects (1) actual variation among taxa and regions (e.g., Benke et al. 1999; Hajiesmaeili et al. 2019), (2) possible variation due to differences in the handling of specimens, such as variation in preservation methods (Johnston and Cunjak 1999), and (3) variation in research focus between particular well-studied taxa and regions, versus those that are understudied. Regarding the latter, our study highlights a highly uneven representation of taxa and geographical regions in the available dataset, which prohibits a reliable statistical analysis of parameter variation across taxa and regions at a global scale, beyond what has been done (e.g., Smock 1980; Schoener 1980; Meyer 1989; Poepperl 1998; Bried and Ervin 2007; Martin et al. 2014). However, visual inspection of the relationships suggests some overarching patterns that may help steer future research priorities.

For instance, based on the currently available data, Coleoptera seems to show more variation in the scaling parameter a than in the exponent parameter b. This implies that the use of a less precise exponent parameter is less critical than the use of a less precise scaling parameter: using exponent parameters at the order level (instead of genus or family level) or from other geographic regions may be relatively more appropriate than using these as a scaling parameter. However, given the great variety of body shapes in the aquatic life stages of the large order of Coleoptera, using the order level or even family level equation may be less appropriate than in other orders. Important for this order is that studies clearly report the life stage of the specimens to avoid confusion since both the larvae and adults are aquatic and exhibit highly different body shapes and sizes. We therefore report the size ranges of the captured specimens for all data, as some parameters may not represent the species in its entire life stage and regressions should only be considered valid for the size ranges on which they are based. Furthermore, variation within larger orders such as Trichoptera (50 families, 605 genera), Odonata (46 families, 670 genera) and Ephemeroptera (44 families, 484 genera) may introduce more variance on order-level estimates than in smaller orders such as Plecoptera (17 families, 310 genera) or Megaloptera/Neuroptera (5 families, 50 genera), although current data availability prohibits testing this assumption. The fact that for Hemiptera there are only very few studies available outside North America may be useful to know a priori for studies that are planning work on the biomass of Hemiptera on other continents (i.e., these should realise that accurate regressions will be difficult to obtain for their study area).

Geographic variation in parameter values seemed particularly strong in Coleoptera, Diptera, Ephemeroptera, and Plecoptera. However, these orders are also the orders with the most data available, providing more power to detect geographic variation. This geographic variation is also expected to be related to temperature, following the temperature–size rule (Atkinson 1995), which states that in ectotherms, higher temperatures can lead to smaller adults or metamorphic body sizes—and may affect growth and consequently morphology. Variation in length–mass relationships within the same taxonomic group can thus also be driven by environmental variation within regions, as a consequence of, for example, altitude or latitude. All in all, our data emphasise that there is a need for (1) more data on variation among families that is collected standardised within the same geographic regions, (2) targeted collections of different orders in the same study areas, to reveal variation among genera, and (3) careful reporting of the exact methodologies used, to allow detecting and mitigating variation introduced due to different methods (e.g., preservation methods, sample sizes, drying and cooling processes). Consortia of researchers from multiple biogeographic regions would be valuable to assess geographic variation at local, regional, and continental scales within the same genera and families.

Concluding, we here provide a digital database with currently available regression parameters that can be applied to existing body length data and may be of use during the design of new studies. The database can be used to identify knowledge gaps and pave the road for additional data collections. It remains important to keep updating the database, as new variations can also arise due to changes in habitat conditions, global change, or species introductions. This is especially true given the low proportions of all taxonomic richness that exist that we have thus far covered at fine taxonomic resolutions.