An estimated 400–800 million tons of prey are annually killed by the global spider community

1Spiders, which evolved from an arachnid ancestor during the Devonian period around 400 million years ago, are among the most common and abundant predators in terrestrial ecosystems (Turnbull; Coddington and Levi; Selden et al. ). For instance, Turnbull calculated an overall mean density of 131 spiders m−2 based on assessments from many different areas of the globe, and Nyffeler found an overall mean density of 152 spiders m−2 for a large variety of grassland habitats. Under favorable conditions, spiders can reach peak densities of up to 1000 individuals m−2 (Ellenberg et al. ). At the present time, >45,000 species of spiders are described and those exhibit a very diverse range of lifestyles and foraging behaviors (Wise; Platnick ). Barth partially attributes the evolutionary success of spiders to the fact that they are equipped with highly developed sensory systems providing individuals with detailed information about potential predators and prey in their surroundings. All spiders are carnivores, feeding predominantly on insects/collembolans and to a lesser extent on other spiders (Nyffeler; Birkhofer and Wolters; Pekár and Toft ). Very rarely nonarthropod prey and sometimes even plant materials are consumed as a supplement to the arthropod diet (Symondson et al.; Foelix; Nyffeler et al. ). Spiders have efficient survival mechanisms given that they are so numerous and widespread. Their capability to survive under extreme conditions and to disperse by ballooning through the atmosphere from place to place on silken threads allowed spiders to colonize a wide variety of different terrestrial habitats. Some spiders can travel distances of up to 30 km in a single day (Thomas et al. ). There is hardly any terrestrial area on this globe where spiders would be missing. “….Spiders exist in the most northern islands of the Arctic, the hottest and most arid of deserts, at the highest altitudes of any living organisms, in the depths of caves, in the intertidal zone of ocean shores, in bogs and ponds, on high, arid moorlands, sand dunes, and flood plains” (Turnbull ). Because of their high abundance and predominantly insectivorous feeding habits, spiders are suspected to be the main predators of insects (Selden ).

2Due to their secretive lifestyle—some species are for example nocturnal or hunt in litter and soil habitats—the predatory activities of the spiders remain largely unnoticed and it is therefore difficult to estimate their impact on prey. To illustrate the impact of spiders as insect predators, two arachnologists—W. S. Bristowe from England and A. L. Turnbull from Canada—previously tried to quantify the food consumption of spiders by means of extrapolations. In his work “A Book of Spiders,” Bristowe estimated that England and Wales are populated by roughly 2.2 × 1012 spiders and that these spiders may kill ≈2.2 × 1014 insects annually. Bristowe went one step further claiming that the weight of insects consumed by the entire British spider fauna would exceed the combined weight of all the humans in Great Britain. Nyffeler conducted a recalculation of Bristowe’s estimate and came to the conclusion that Bristowe probably overestimated the overall prey kill of the spiders since Great Britain consists predominantly of agricultural land characterized by reduced annual prey consumption. Turnbull, on the other hand, estimated that the average total weight of food annually consumed by spiders would amount to 4.25 × 103 metric tons km−2 land area. This latter value is of the same magnitude as the net primary production in terrestrial ecosystems, which is irreconcilable with ecological theory (Nyffeler ).

3Here, we provide estimates for the standing biomass of the global spider community and the annual biomass of prey that is killed by the global spider community in individual biome types and worldwide based on literature data.

4A total of 65 values of spider biomass m−2 were gathered from the literature. The data were assigned to the following seven groups of terrestrial biomes: (1) tropical forests, (2) temperate and boreal forests, (3) tropical grasslands and savannas, (4) temperate grasslands (incl. old fields, permanent pastures, mown meadows) and Mediterranean shrublands, (5) annual cropland, (6) deserts, and (7) Arctic tundra. To retrieve comparable data, all values were converted to fresh weight m−2 taking into account an average water content of the spider body of ≈75% (Pulz ). The data were pooled by computing an average biomass value (g m−2) for each biome type. By extrapolation—using the global land cover data from Saugier et al.—the standing biomass of the global spider community was then computed (Table 1).

5We used simple models involving few assumptions as is advised in cases where a field of study is still largely undeveloped (Weathers and Weathers ). Two different approaches were taken to estimate the annual prey kill of the global spider community. In the case of method I, the estimate is based on the spiders’ food requirements per unit body weight known from the literature in combination with spider biomass m−2 values (data for various biome types being taken from the literature), whereas method II is based on complete assessments of the spiders’ annual prey kill (e.g., prey censuses in the field combined with web density estimates) in selected biome types published in the literature. The two estimation methods are based on different sets of studies (with zero overlap of data between the two methods).

6Prey items that are killed in webs but remain uneaten are considered prey as well (Nentwig ). This issue is playing a role when dealing with spider communities dominated by large-sized orb-weaving spiders known to often kill prey in excess (“wasteful killing”). Accordingly, authors who conducted prey censuses of large-sized orb weavers usually have taken into account the “prey killed in webs but uneaten” in their assessments (see Robinson and Robinson; Kajak et al.; Nyffeler,; Nyffeler and Benz,,; Malt ). In other studies dealing predominantly with cursorial hunters and/or small-sized web-building spiders which rarely catch prey in excess under natural conditions (Nyffeler, pers. obs.), this issue was disregarded.

7Method I: Based on the spider biomass m−2 values from Table 1 and an assumptive food intake rate (mg prey per mg spider body mass day−1), the prey kill m−2 day−1 of spider communities for each of the seven biome types was computed. These values were extrapolated to prey kill m−2 year−1 for each biome type, considering the length of the spiders’ feeding season (in days, see assumptions below). By multiplying the prey kill m−2 year−1 with the corresponding area size of a particular biome type (based on Saugier et al. ), a prey kill subtotal was derived for each biome type. Summing up the seven subtotals produced an estimate of the annual prey kill of the global spider community (Table 2). The estimate derived by method I was based on the following assumptions:

8- Assumption 1: Spiders have pulsed feeding patterns, with periods of excessive feeding (when food is very abundant) alternating with episodes of starvation (when prey gets scarce or spiders are inactive) (Turnbull; Anderson ). During periods of high feeding activity, the spiders store surplus energy in their body’s interstitial tissue as lipid or glycogen (Foelix ). The spiders depend on these stored energy reserves during periods of starvation (e.g., on rainy days). To estimate the annual prey kill of the global spider community, we proceed from an overall mean food intake which is intermediate between pronounced high or low daily food consumption. After an extensive literature survey on spider feeding, we propose an average daily food ingestion rate of ≈0.1 mg per milligram spider body mass which is the equivalent to ≈10% of a spider’s body weight (all values expressed as fresh weight). A daily food intake in this order of magnitude appears to be typical of most species of free-living araneomorph spiders in forests, grasslands, and agroecosystems (see Edgar; Robinson and Robinson; Van Hook; Foelix; Nyffeler, pers. obs.) except a few rare cases of extraordinarily low food intake (e.g., Santana et al.; Henschel ). In the case of desert spiders, a daily food ingestion rate of 0.01–0.04 mg per milligram spider body mass was used (see Lubin and Henschel; Henschel ). Incidents of wasteful killing (and coupled with it “partial consumption”) in cursorial spiders apparently occur very rarely under natural conditions (Nyffeler, pers. obs.), and this issue has therefore not been taken into account in this study.

9- Assumption 2: Spiders ingest, on average, ≈80% of the biomass of a killed prey (Edgar; Moulder and Reichle ). Hence, we proceed with the assumption that the daily prey kill equals the daily amount of food ingested multiplied by the factor 1.25.

10- Assumption 3: We assume that spiders forage on 365 days year−1 in tropical forests, on 240 days year−1 in deserts, on 180 days year−1 in temperate forests as well as tropical and temperate grasslands, on 120 days year−1 in the arctic tundra, and on 60–130 days year−1 in annual cropland (Kajak et al.; Robinson and Robinson,,; Breymeyer; Shook; Nyffeler; Byzova et al. ). The contribution of winter-active spiders in terms of prey kill in temperate and cold climates (see Aitchison ) is considered to be very low and has therefore been neglected.

11- Assumption 4: Spider biomass in forests has in most cases been assessed with the Berlese-Tullgren funnel method. This technique is limited to the investigation of spiders on the forest floor, and the calculated biomass values underestimate true biomasses. In temperate forests, at least 20% of the spider biomass are found in the canopy and understory (see Turnbull; Reichle and Crossley; Moulder and Reichle; Zitnanska ). This pattern seems to hold for tropical forests (see Basset et al.; Silva; Yanoviak et al.; Ellwood and Foster; Dial et al. ). By multiplying the litter spider biomass values with a correction factor of 1.25, estimates for the total spider biomass in temperate, boreal, and tropical forests were obtained (Table 1).

12- Assumption 5: For biomass m−2 of spiders associated with Mediterranean shrublands, no data are available. We arbitrarily placed this biome type in the category “Temperate grasslands (old fields, permanent pastures, mown meadows)”. The area size of Mediterranean shrublands is small (2.8 × 1012 m2) relative to the global terrestrial area, and a possible error resulting from insufficient data can be considered to be negligible.

13Method II: The second approach is based on published studies of the annual prey kill of spider communities in various biome types (see Kirchner; Reichle and Crossley; Kajak et al.; Van Hook; Moulder and Reichle; Robinson and Robinson; Luczak; Nyffeler,; Nyffeler and Benz,,,,; Schaefer; Ysnel; Jmhasly and Nentwig; Malt ). For purposes of comparison, all prey kill values (including those expressed in terms of energy flow) were converted to grams of fresh weight m−2 year−1. Values were converted taking into account a prey water content of ≈75% (see Hagstrum; Edgar ) and a caloric equivalent of prey of ≈23.5 kJ g−1 dry weight (mean value from literature data, see Hagstrum; Moulder and Reichle ). Thus, 1 g fresh weight prey biomass equals ≈5.875 kJ. By means of extrapolation, the annual prey kill of the global spider community was computed, taking into account the global coverage of the different biome types. The global annual prey kill assessment with method II was based on the following assumptions:

14- Assumption 1: Assessments of the annual prey kill of spider communities in tropical forests are currently unavailable. In lieu thereof, a study by Robinson and Robinson on the prey kill by the web-building spider community of an insecticide-free coffee plantation in New Guinea was used as a surrogate. We were operating on the assumption that the spider communities of tropical insecticide-free coffee plantations are to some degree comparable to those of tropical forests, given that coffee plantations are inhabited to a large extent by tropical woodland spiders (e.g., Nephila maculata) (Robinson and Robinson,; Robinson et al.; Lubin ). Robinson and Robinson came to the conclusion that the web-building spider community in their study killed 16 g insect prey m−2 year−1. These authors suggested that the annual prey kill may even have been twice as high (≈32 g insect prey m−2 year−1) if hunting spiders would have been considered. The spider density in this coffee plantation (5.8 individuals m−2) was higher than the reported densities in the understory of tropical rain forests (3.3–3.6 individuals m−2; Rypstra; Reagan and Waide ). However, we have to take into account that coffee plants reach a height of only 3–3.5 m, whereas tropical forest trees grow to a height of up to 55 m (Silva ). The canopy of tropical forests is inhabited by an abundant spider fauna (Basset et al.; Russell-Smith and Stork; Silva; Ellwood and Foster ), and it is to be expected that those spiders kill considerable numbers of insects in addition to the insects killed by the spiders of the understory. Thus, it is well possible that the annual prey kill by spiders in tropical forests does exceed the conservative estimate of 16 g insect prey m−2 year−1.

15- Assumption 2: Annual prey kill values for temperate forests appear to vary widely. Kirchner estimated an annual prey kill of ≈10 g m−2 year−1 for a semi-natural temperate forest in Central Europe, whereas lower values were reported for managed temperate forests. The annual prey kill in managed temperate deciduous forests in North America and Central Europe was estimated at ≈2 g m−2 year−1 (calculated by combining data for the spiders of the forest floor, understory, and canopy [Reichle and Crossley; Moulder and Reichle; Schaefer ]).

16- Assumption 3: Annual prey kill values for unmanaged grasslands vary widely from ≈2 g m−2 year−1 (Ysnel ) up to >10 g m−2 year−1 (Kajak et al.; Nyffeler and Benz ). In order to avoid overestimation, we used a conservative annual prey kill range of 2–10 g m−2 year−1 for grasslands and savannas (Table 3). The split in global coverage between unmanaged grassland/savannas (13.7 × 1012 m2) vs. permanent pastures/mown meadows (28.9 × 1012 m2) was based on SAGE/GTAP data http://www.agter.org/images/merlet_c2a_cultivablelands_G1.png

17- Assumption 4: Published prey kill records for desert spider or arctic tundra communities are not available. Nevertheless, based on published natural history data (Polis; Henschel ), we conclude that the annual prey kill in deserts is most likely very low. One might compare deserts to some degree to urban environments where the annual prey kill by spiders is also very low (0.2 g m−2 year−1; Nyffeler ). Supposing that the annual prey kill in deserts might be equally low as in urban areas, we arbitrarily assigned an assumptive value of ≈0.2 g m−2 year−1 to the spider communities in desert areas in order to be able to compute the global annual prey kill with method II. Due to many similarities between tundra and agricultural habitats (regarding population densities, body size composition and faunistic composition), we assume that the annual prey kill in arctic tundra sites might be comparable in magnitude to field crops in Europe (≈0.1–1 g m−2 year−1; Table 3).

18Summing up the estimated prey kill subtotals for the seven biome types produced a second estimate of the annual prey kill of the global spider community.

19Based on estimates of the average spider biomass m−2 in various terrestrial biomes, extrapolation suggests that the total standing biomass of the global spider community equals 25.09 × 1012 g (= 25 million metric tons fresh weight; Table 1). Biomass estimates m−2 follow the order forests > grasslands/shrublands > croplands, deserts, and tundra, which also reflects the order of total spider biomass per biome type worldwide.

20The calculation of the annual prey kill by the global spider community with method I resulted in an estimate of 721 × 1012 g year−1 (= roughly 700 million tons year−1; Table 2). To derive an estimate to which degree reduced feeding activity during rainy days would affect the global annual prey kill, we recalculated this estimate assuming that it rained during one third of the feeding season with no prey being captured on rainy days. This simple scenario still leads to a global annual prey kill of 460 million tons year−1. Estimates derived from method I therefore suggest that the annual prey kill of the global spider community may be in the range of 460–700 million tons year−1. Our assessment with method II produced an estimate of the global annual prey kill of 395–805 × 1012 g year−1 (= roughly 400–800 million tons year−1; Table 3). The estimates computed with the two methods are highly comparable in magnitude (Tables 2 and 3). Together, the two different methods suggest that the global annual prey kill is presumably in the range of 400–800 million tons year−1.

21Our extrapolations resulted in an estimated annual prey kill by the global spider community in the range of 400–800 million tons year−1 (Tables 2 and 3), which equals ≈1‰ of the global terrestrial net primary production (see Vitousek et al. ). For comparison, the human world population does consume an estimated 400 million tons of meat and fish annually (Bruinsma; Food and Agriculture Organization of the United Nations ). Furthermore, our estimates for spiders appear to be of the same order of magnitude as the prey kill by whales (Cetacea) in the world’s oceans which has been estimated to be in the range of 280–500 million tons annually (Yodzis ). By contrast, the annual food consumption of all the world’s seabirds is estimated at 70 million tons (Brooke ).

22Overall, fewer than 10 arthropod orders (Diptera, Hemiptera, Hymenoptera, Collembola, Coleoptera, Lepidoptera, Orthoptera, and Araneae) make up the majority of the prey of spiders (e.g., Turnbull; Kajak et al.,; Robinson and Robinson,,; Nyffeler, ). Apart from insects and collembolans, spiders are another important component in spider diets (“intraguild predation” sensu Polis et al. ). Spider communities in the warmer areas (often dominated by cursorial hunters) have a higher percentage of spiders in their diets than spider communities in the colder climates (see Nyffeler; Nyffeler and Sunderland ) and cursorial spiders feed more frequently on other spiders than web-builders (Birkhofer and Wolters ). Spiders feed on all stages (eggs, immatures, and adults) of their arthropod prey (Nyffeler et al. ). In addition to this, some larger spiders occasionally prey on earthworms, slugs, snails, and small vertebrates (Foelix ). Also, spiders have been reported supplementing their animal diet by feeding on plant materials (Nyffeler et al. ).

23Spiders in forests and grasslands accounted for more than 95% of the annual prey kill of the global spider community (i.e., 697 million tons year−1 estimated with method I and 387–784 million tons year−1 with method II), whereas spiders in other biomes (i.e., annual crop fields, urban areas, deserts, arctic tundra) were less significant contributors to the global prey kill (24 million tons year−1 estimated with method I and 8–21 million tons year−1 with method II, Tables 2 and 3). Forests, grasslands, and savannas cover an area of 87 million km2, which is about two thirds of the global terrestrial surface area (Saugier et al. ). Forests, grasslands, and savannas are less frequently disturbed than, e.g., agricultural or urban areas and allow spider populations to build up a higher biomass (Table 1). Both these differences explain the high prey kill in these biomes (Tables 2 and 3). These spiders further feed on many forest and grassland pest species underlining the important role of spiders as providers of biological control services (e.g., Juillet; Kirchner,; Pointing; Kajak et al.,; Van Hook; Jennings and Pase; Furuta; Schmitz; Oedekoven and Joern ). However, spiders do not only kill pest prey but also consume other beneficial arthropods (including large numbers of honey bees) in forest and grassland habitats (Nyffeler and Benz; Malt ).

24In relation to pest control, it is noteworthy that the spiders associated with annual crops only contribute less than 2% to the global annual prey kill. This can be explained by the fact that annual crop fields are “disturbed habitats” characterized by low spider biomass and a relatively short feeding season (Luczak; Nyffeler and Benz; Nyffeler et al. ). Nevertheless, in wheat-, rice-, and cotton-growing areas with no or very low pesticide usage, the presence of spiders (in combination with other predators) may at times have a beneficial effect in slowing down the population growth of hemipteran pests (Kiritani et al.; Sunderland et al.; Nyffeler and Benz,,; Nyffeler et al.,,; Jmhasly and Nentwig; Birkhofer et al., ).

25Desert spiders only account for a small percentage (≤2%) of the global annual prey kill, but deserts cover a vast area (27.7 million km2) of the globe (Saugier et al. ). Due to adverse environmental conditions, prey availability in deserts is very low and these biomes are often populated by spiders in very low densities (Shook; Polis; Lubin and Henschel; Henschel ). There are exceptions to this trend. Polis and Hurd described small desert islands in the Gulf of California, Mexico, where spiders occur in extraordinarily high numbers as a consequence of allochthonous energy input from the ocean. But the area of such island deserts is so small compared to the globe’s total desert area that its contribution to the global spider prey kill must be considered as negligible. Despite the fact that prey biomass killed m−2 by desert spiders appears to be relatively low compared to spider communities in other biomes, these arachnids are noticeable top predators in desert food webs (Polis and McCormick; Polis; Polis and Yamashita ). The relative contribution of spiders of the Arctic tundra to the global annual prey kill is equally low as for desert-living spider communities (<1%). The climatic seasonality and short foraging periods in this biome certainly contribute to those low estimates. However, spiders in the Arctic tundra (dominated by individuals from the family Linyphiidae) play an important ecological role by entrapping in their webs nutrients that originated from wind-blown allochthonous inputs (mostly chironomid midges). These nutrients are thereafter retained within the system and contribute to early stages of ecosystem development (Hodkinson et al. ).

26Our estimates of the global annual prey kill (Tables 2 and 3) imply that spiders exert considerable predation pressure on insect populations, especially in forests and grasslands. These estimates are supported by exclusion studies in forest and grassland habitats in different parts of the world (Clarke and Grant; Kajak et al.; Schmitz; Oedekoven and Joern; Lawrence and Wise; Tanhuanpää et al. ). After spiders had been manually removed from experimental exclosure plots, a significant increase of the number of insects and collembolans was noticed compared to control plots with spiders. Greenstone, however, stated that even though spiders can achieve very high densities and consume large quantities of insect prey, they may not necessarily have a significant role in the regulation of insect populations. Spiders depress herbivorous insect populations by 1–20% (Juillet; Pointing; Kirchner; Furuta; Nakamura and Nakamura; Oedekoven and Joern; Tanhuanpää et al. ). Acting in concert with other natural enemies (ants, ground beetles, predaceous bugs, birds, etc.), spider communities certainly exert significant predation pressure on herbivorous insect populations (e.g., Kajak et al. ).

27There is also indirect support for the significance of spider predation in terrestrial biomes: a large number of insect species have evolved morphological and/or behavioral adaptations to reduce the risk of predation by spiders (see Eisner et al.; Wise ). This concerns in particular the adults of ten thousands of species of moths, butterflies, and skippers (Lepidoptera) which evolved an escape mechanism allowing them to disentangle themselves from the webs of ecribellate spiders (see Eisner et al. ). Eisner et al. stated “….Moths, by virtue of the loose scales that cover their wings and bodies, are admirably adapted to elude capture by orb-weaving spiders. Rather than sticking to the web, these may simply lose some of their scales to the viscid threads, and then fly on.”

28Further indirect evidence for the enormous ecological significance of spiders is that they serve as a food source for an incredibly diverse complex of arthropod-eating carnivores and, given our estimated global spider biomass (25 million metric tons fresh weight), spiders certainly are a crucial source of nutrition for many predator species. An estimated 8000–10,000 species of highly specialized predators/parasitoids/parasites feed exclusively on spiders, all of them being evolutionarily adapted to this special lifestyle (Table 4). In addition to this, an estimated 3000–5000 polyphagous bird species include spiders in their diets. Many of these bird species feed heavily on spiders in certain periods (i.e., spiders constituting 20–95% of the birds’ prey biomass [Kuitunen and Törmälä; Poulin and Lefebvre; Naef-Daenzer et al. ]). Likewise, many species of frogs, toads, salamanders, newts, lizards, snakes, shrews, mice, rats, and bats include spiders in their diets (Shine; Schulz and Wainer; Van Sluys and Rocha; Marques et al. ). Even fish from more than 20 families have been reported consuming spiders that mistakenly landed on the water surface after ballooning across lakes or that fell into water from overhanging trees (e.g., Bristowe; David and Closs ).

29In recent years, researchers have shown experimentally that spiders not only affect insects by inflicting mortality on them, they additionally have indirect (nonlethal) effects by intimidating them to the point where insects reduce their feeding activity in the presence of spiders in order to decrease predation risk (Schmitz et al.; Schmitz; Snyder and Wise; Hlivko and Rypstra ). Such nonlethal effects of spiders can result in reduced herbivore damage. Hlivko and Rypstra noted “….the potential impact of spiders may be underestimated in food web studies that only consider predation rates….” Thus, the spiders’ huge predation impact documented in our study (Tables 2 and 3) is even further enhanced due to additional nonlethal effects.

30The estimated standing biomass of the global spider community is impressive (≈25 million tons, Table 1). There are few groups of terrestrial predaceous arthropods that can compare with spiders in terms of abundance and biomass. An exception are ants (Formicidae estimated at approx. 280 million tons, Hölldobler and Wilson ), but most species of ants are omnivorous and utilize a much broader range of food resources including considerable amounts of plant materials (Hölldobler and Wilson ). Here, we suggest that the predation impact of spider communities is particularly high in forest and grassland biomes, with considerably lower impact in desert, urban, and tundra biomes. These estimates emphasize the important role that spider predation plays in semi-natural and natural habitats, as many economically important pests and disease vectors breed in those forest and grassland biomes. We hope that these estimates and their significant magnitude raise public awareness and increase the level of appreciation for the important global role of spiders in terrestrial food webs.