Assessing the Response of Different Soil Arthropod Communities to Fire: A Case Study from Northwestern Africa

: In recent decades, forest ﬁres in the Mediterranean basin have been increasing in frequency, intensity, and the area burnt. Simultaneously, insects, a group with extraordinary biodiversity that provides vital ecosystem services such as pollination and decomposition, are undergoing a precipitous decline. Unfortunately, the impact of ﬁre on arthropod communities has been poorly addressed despite the high diversity of taxonomic and functional arthropod groups. Responses to ﬁre can differ considerably, depending on the life history and functional traits of the species. In the present study, we investigate the short-term impact of ﬁre (three years after a blaze) on the abundance and species composition of soil arthropods in a burnt pine forest located in Ceuta (Spain, northwestern Africa). Soil arthropods were collected from pitfall traps in burnt and unburnt pine forest sampling points. In terms of total abundance per taxonomic order, Blattodea and Diptera were the only orders seemingly affected by the ﬁre, whereas other arthropod groups (e.g., Araneae, Coleoptera, and Isopoda) showed no differences. In terms of species composition, Coleoptera and Formicidae (Hymenoptera) communities differed between burnt and unburnt sampling points, having more species associated with burnt areas than with unburnt ones. In burnt areas, some species from open areas built nests, fed in/on the ground, and dispersed over longer distances. Within the unburnt plots, we found more species in vegetated habitats, particularly those with shorter dispersal distances. We conclude that arthropod communities differ between burnt and unburnt sites and that the response of each taxon appears to be related to particular functional traits such as habitat preference (from open to forested landscapes) and ecological specialization (from generalist to specialist species).


Introduction
Fire, one of the most influential natural disturbances in fire-prone areas such as the Mediterranean region [1], can act as a major driver of animal community composition [2]. Over the last few decades, humans have been altering fire regimes through land-use shifts (land abandonment and forestry) as well as human contributions to climate change and its consequences, particularly drought [3]. Even though fire is commonly viewed as disastrous, it is considered a critical driver of biodiversity in fire-prone ecosystems [4,5], accelerating

Arthropod Sampling
Soil arthropods were sampled in a pine plantation in which 60 ha burned during a summer fire in 2019. Gutter pitfall traps (6 cm in diameter) were used, partly filled with soapy water and salt. Five sampling points were established: three in the unburnt and two in the burnt area. In the latter, pitfalls were set in patches with similar pre-fire vegetation (i.e., pine forest) to that of the unburnt area. We acknowledge that the number of replicates is small; this was caused by the small size of the fire perimeter and the lack of many similar pre-fire vegetation patches (pine plantations) in areas surrounding the burnt area. Unburnt pine plots far from a burnt area could demonstrate that the differences between unburnt and burnt plots are caused by other factors (e.g., climate, slope and aspect, and lithology) than fire.
At each sampling point, 5 pitfalls were placed 10 m apart along 50 m linear transects. After pitfall installation, the soil was restored with similar litter from the surrounding area. The distance between burnt and unburnt sampling points averaged 104 ± 2.46 (standard error) meters. Pitfalls were collected after 6 consecutive days on two sampling occasions (July 2022 and October 2022). No pitfalls became completely dry after the 6-day period. Arthropods from pitfall traps were preserved in 70% ethanol until identification. In the lab, specimens were separated by class and identified to the order level. Formicidae (Hymenoptera), Coleoptera, and Orthoptera specimens were classified to species level, thanks to the taxonomic and faunistic expertise of northwestern African fauna by the authors (P.B. for Orthoptera, J.L.R. for Coleoptera, and A.T. for Formicidae). Arthropod sampling was conducted under permit from the Ceuta authorities.

Data Analysis
Three types of analyses were performed: (1) Abundance values of the top 10 most-abundant arthropod orders were compared between pitfalls in burnt and unburnt areas, using Generalized Linear Mixed Models (GLMMs). This analysis was carried out using a Poisson distribution due to the discrete nature of the dependent variables (counting the number of individuals captured). The fire condition (burnt or unburnt) of the sampling points was used as an independent variable, whereas the sampling point and the sampling month (July and October) were treated as random effects. Statistical analyses were performed using the lme4 package [42], and figures were performed using the ggplot 2 package [43]. (2) The composition and species abundance of ant, beetle, and orthopteran communities found in the burnt vs. unburnt pitfall traps were compared by permutational analysis of variance (PERMANOVA). For each arthropod group, the pairwise similarity in species abundance and presence among pitfall traps were assessed using the Bray-Curtis similarity distance for the relative abundance data and using the adonis2 function from the vegan package [44]. Similar to the GLMM design, the fire condition was used as a fixed factor, whereas the sampling point and season were used as random factors in the PERMANOVAs. All analyses were performed using R software (Core Team 2021). (3) Based on the abundance of beetle, ant, and grasshopper species per pitfall trap, we calculated the Shannon diversity index. Then, we used Generalized Linear Mixed Models (GLMMs) using the lme4 package [42], with a Gaussian distribution, to examine the effect of the fire condition (burnt and unburnt) on the Shannon diversity per trap. The sampling point and the sampling month (July and October) were treated as random effects.

Results
In total, 3131 arthropods were collected, with a mean of 67 arthropods per pitfall (±7.57 SE) belonging to four classes, namely, Crustacea (n = 233 individuals), Arachnida (n = 966 individuals), Myriapoda (n = 3 individuals), and Insecta (n = 1963 individuals) ( Table 1). Among insects, the most-abundant taxa collected were Formicidae (Hymenoptera). The GLMMs indicated that the total arthropod abundance values per pitfall did not differ significantly between burnt and unburnt samples ( Table 2). The comparison for the top 10 most-abundant arthropod orders (Araneae, Coleoptera, Isopoda, Blattodea, Diptera, Collembola, Orthoptera, Formicidae (Hymenoptera), and non-Formicidae Hymenoptera) and the subclass Acari did not show differences in abundance per pitfall between burnt and unburnt samples for the majority of groups, except in the cases of Blattodea and Diptera (Table 2). For these two orders, the abundance was higher in pitfall traps from unburnt sampling points than from burnt sampling points (Table 2 and Figure 2a  We recorded a total of 937 ants (Formicidae) from 16 species, 185 individuals from 35 species of Coleoptera belonging to 19 families, and a total of 11 Orthoptera from 5 species (Supplementary Materials Table S1). The PERMANOVA analyses revealed that the beetle and ant communities differed between burnt and unburnt sampling points (Table 3). By contrast, the grasshopper species composition did not differ in this regard, probably due to the small sample size collected for this group. Overall (with beetle, ant, and grasshopper species pooled), no differences were observed in the Shannon diversity scores between burnt and unburnt sampling points (t = 0.012, p = 0.990; Figure 3).
Tenebrionidae (seven mainly saprophagous species), and Carabidae (six species of predatory and phytophagous species including a granivorous species, Carterus rotundicollis, present only at burnt sites) were the beetle families with the highest numbers of species collected. Among the 35 species of beetles recorded (Supplementary Materials Table S1), 16 were found exclusively in burnt areas, 10 only in unburnt areas, and 9 in both areas.
Among the species exclusively found at unburnt sites, at least four showed an affinity for moist substrates or were clearly hygrophilous (Brachinus andalusicus, Dienerella separanda, Ocypus olens, and Sepedophilus sp.), while another species, Ripidius quadriceps, which is very rare in North Africa, parasitizes species of Blattodea (mainly genera Ectobius and Blattella [53,54]), a taxonomic group showing a significantly higher abundance in unburnt areas.
For ants, two subfamilies were recorded: Myrmecinae, with 10 species and five genera, and Formicidae, with 6 species and three genera (Supplementary Materials Table S1). Of the 16 species captured (all native to Ceuta), 9 were forest species (mesoforest or euforest species), and 7 had a wide range of habitats, usually nesting in open and well-exposed places. Of these nine forest species, four species were common in burnt and unburnt areas, whereas three species were located only in burnt areas. These three species were Camponotus spissinodis, which is a strictly arboreal species that burrows into and nests in wood; Camponotus ali, which lives under stones; and Plagiolepis barbara, which lives under stones and in litter. The rest of the species found in the burnt areas (Messor barbarus, M. sanctus, and Cataglyphis viatica) were species of open and thermophilic areas.
The forest species Temnothorax recedens, Crematogaster scutellaris, and Tetramorium exasperatum were collected exclusively in unburnt areas (Supplementary Materials Table S1). T. recedens, a relatively thermophilic mesoforest species, was found mostly in subhumid, warm, and mild climate zones [56], C. scutellaris nests in dead branches, sometimes spreading to the ground at the base of a tree [57], while T. exasperatum not only nests in litter and under stones in cork oak forests [58,59] but also occurs in open places [60].
Tetramorium semilaeve proved to be the most-abundant ant species in burnt as well as unburnt areas. The species is known to use a wide range of open and well-exposed habitats such as grasslands and scrublands [60]. It was followed by two forest species, Temnothorax curtulus and Tetramorium exasperatum, in unburnt areas, and by two species of open and dry habitats, Crematogaster auberti and Messor sanctus, in burnt areas [60,61].
Orthoptera captures were scarce (Supplementary Materials Table S1) and mostly of immature specimens. In general, they were generalist species. Pezotettix giornae appeared in herbaceous environments, quite commonly at medium altitudes and coastal plains, reaching up to 2300 m a.s.l., though it was absent near the sea. It, reportedly, frequently occurs along with Calliptamus barbarus on fallow lands [62]. This latter species is also common in flat areas in wastelands and in open scrub surrounded by fallow areas. It is distributed from the coast to elevations above 2000 m. C. barbarus is the most-frequent and abundant Orthoptera species in northern Morocco, with adults appearing from June to October [63]. Existing records of Gryllomorpha uclensis in North Africa should be reviewed, as several very morphologically similar species were identified to the species level by studying only the male genitalia [64,65]. Its presence was confirmed in Algeria [66]. In our case, no adult males were captured to allow for a study of this structure, so the identification is preliminary until there is confirmation with adult male specimens.

Discussion
This is the first study available that characterizes the species-level effects of fire on several soil arthropod groups in northwestern Africa. The present field study data show that the short-term arthropod-community responses to fire differed among arthropod groups. On the whole, arthropod abundance was indistinguishable between burnt and unburnt areas, with the exceptions of Blattodea and Diptera, which decreased in burnt areas. However, a finer analysis in terms of species composition revealed that beetle and ant communities varied between burnt and unburnt sampling points. Although the sampling design was limited to a single fire, several pieces of evidence indicated that arthropod communities differ between burnt and unburnt points and that species responses are linked in part to their functional traits [23].

Effects of Fire on the Abundance of Arthropod Taxa
Landscape openness is a major driver affecting the soil arthropod fauna over the short-term following a fire [67]. However, the response to fire can vary among arthropod taxa, as particular traits can boost the probability of surviving fire, the capacity for dealing with future environmental conditions, and the likelihood of colonizing burnt areas [68,69]. In terms of total abundance, the results show that Blattodea and Diptera were the only orders apparently affected by fire. These groups are known to be among those sensitive to fire [70]. A possible explanation for the negative effect of fire on these two orders is the short-term elimination of the existing layer of litter and decomposing vegetation on the ground. Thus, the decreases for Blattodea and Diptera could reflect an intolerance to soil dryness in recently burnt habitats [71]. Frouz [72] found that soil moisture and the input of organic matter are the main factors that influence Diptera abundance after fire. The decline of Blattodea in burnt areas might be caused by the post-fire alteration of the vegetation structure in pine plantations [73] and a short-term post-fire reduction in leaf-litter deposition, the primary food source for cockroaches [30]. While these two orders decreased after fire, other groups of arthropods (i.e., Araneae, Coleoptera, and Isopoda) showed significant resilience in terms of abundance. This is most probably attributable to the depth at which they live during their preimaginal stages, their burrowing capability, the availability of shelters such as logs and rock piles [31], and their high capacity for early colonization from nearby unburnt areas, especially in open habitats [74][75][76].

Effects of Fire on the Composition of Arthropod Communities
On a short-term basis, fire is an environmental filter that screens for species that are better-suited to the restricted environmental conditions following a blaze [77]. The early post-fire succession enhances open areas and promotes a change in dominant animal species, which frequently results in different assemblages in burnt and unburnt areas [78,79]. Within the burnt plots, species more characteristically from open areas were expected to be found, including those that construct nests, forage in/on the soil, and disperse over longer distances. For unburnt plots, species more typically found in vegetated habitats were expected, especially those with shorter dispersal distances [10]. The results support these expectations, as significant differences were found in the (functional) composition of ant and beetle communities between burnt and unburnt sites.
The most-abundant beetle species in both areas were flightless, generalists in terms of habitat preference, and opportunistic (except Pactolinus major, which also takes refuge under stones [50]), so their capacity for dispersal and the rapid recolonization of burnt areas would presumably be lower than that for flying species. The abundance of these species at burnt and unburnt sites proved similar. This is because they all present hypogeous preimaginal stages, with larvae and pupae living buried in the ground at a certain depth (Tenebrionidae and Geotrupidae) and, in some cases (Carabidae), also under stones (which act as shelter), giving them a high resilience to fire, especially against more superficial burning [76,80].
At burnt points, generalist, thermophilic, and heliophilous beetle taxa predominated, preferring open habitats or those with little tree cover. The greater richness of the beetle species found in burnt areas was widely observed by other authors (e.g., [74,75,[81][82][83]). In this case, the pattern would have two main causes (see [76]): the resistance and resilience to fire of generalist species and the relatively rapid exogenous colonization from nearby non-forested areas by thermophilic species, typical of open habitats with scrub. The pine plantations in the Ceuta area are generally homogeneous and have shady undergrowth with little diversity [38, authors' personal observation)]. Three years post-fire (when the sampling was made), burnt pine patches still lacked tree cover and were, at that time, being colonized by diverse scrub and herbaceous plants; furthermore, these patches presented greater floristic diversity, structural complexity, and heterogeneity than the unburnt pine stands did. This structure offers greater availability of microhabitats and trophic resources for coleopterans, especially phytophagous species (including canthophilous), saprophagous species, and their predators [75,84,85]. In addition, the reduction or disappearance of tree cover would favor the colonization of heliophilous and floricolous beetles, typical of open habitats [74,79,85].
Most studies reported that ant composition is significantly altered by fire [86][87][88]. As such, fire is considered a main driver of functional diversity change in ants [68]. Our findings show that more ant species were associated with burnt areas. This could be explained by the fact that the majority of ant species nest in the soil and dig galleries that are sometimes very deep, with a slight temperature rise a few centimeters below ground, ensuring that the majority of ants would survive a fire [88][89][90]. The presence of forest species in burnt areas in our study is due to their resistance to fire in cryptic habitats, as in the cases of Camponotus ali and Plagiolepis barbara, which can dig deep galleries under stones, for example, or to the reoccupation of burnt areas from the neighboring forests.
The two harvester ant species (M. barbarus and M. sanctus) are strictly granivorous (the only herbivorous species collected in this study). These two species prefer to nest in open and warm environments rather than under tree canopies [91]. Thus, burnt areas offer them suitable environmental conditions: landscape openness with grass seeds and glumes for food [92]. These post-fire open and thermophilic environments could also encourage the colonization of other species captured only by pitfalls in burnt areas, such as C. viatica.
The presence of other common ant species in both burnt and unburnt pitfalls (e.g., Tetramorium caespitum) evidences their relatively high plasticity, with respect to their habitats and microhabitats (with the exception of Camponotus ruber, which is arboreal). These generalist species can all nest on the ground, under dead spindles, under bark, in lodges, and also in galleries under stones. We presume that this plasticity allows them to escape the direct effects of fire-heat, dryness, and smoke [93].
The three arboreal species collected in this study (Camponotus spissinodis, Crematogaster scutellaris, and Temnothorax recedens) are relatively indicative of the nature of the ecosystem prior to pine reforestation and before the fire. In fact, these species generally characterize oak forests in Morocco ( [56,94], unpublished personal data).
Our findings suggest that the differences in the ant-species composition between burnt and unburnt areas may be due to changes in microclimatic conditions and food resources caused by fire [87]. Moreover, the majority of species found in unburnt areas have relatively large colony sizes (Messor, Camponotus, Cataglyphis, and Solenopsis genera, for example). It was demonstrated that the larger an ant colony is, the better it can buffer disturbances, offering the species more ecological advantages of resistance [68,95]. In addition, a greater abundance was observed in burnt areas for some species associated with the highest values of body size and worker polymorphism traits (the genera Messor, Cataglyphis, and Camponotus). A positive relationship between heat tolerance and body size in ants was supported by several studies [96,97]. Large workers can remain active and move greater distances at higher temperatures than small workers can, as longer legs enable them to elevate themselves over the warm substrate and attain higher running speeds [98].
Several authors indicated that Orthoptera communities have positive responses to fire due to the increase in plant diversity after fire [22,35,99], although this diversity could decrease after several years with vegetal homogenization. Moreover, Orthoptera diversity may vary depending on the season when the fire occurs [35]. Additionally, the depth at which orthopteran egg pods are laid determines their survival after a fire [100]. This situation may also apply to buried crickets. However, the extremely scarce number of Orthoptera captures in our study precludes any conclusions concerning this group. Both caeliferan species, P. giornae and C. barbarus, are common, widely distributed, and also found in agricultural areas [101], whereas cricket species were found to live buried in soil or under large-to medium-sized stones. All these traits suggest that these species can survive brief fire events (the 2019 fire burned for approximately 14 h).

Concluding Remarks
Being by far the richest group of animals in the world in terms of biodiversity, insects need further study to address their conservation needs [10]. For example, scientists have limited knowledge of how insect communities respond to wildfire, as the pyroentomological literature presents a mixture of both positive and negative responses [102]. The ability to forecast ecological and community responses to wildfire necessitates an understanding of how fire and biota relate to each other [103]. In this light, the present research focuses on the species-level effects of fire on several soil arthropod groups. Despite the small scale of the fire event studied, compared to the current megafires in the Mediterranean basin [104,105], the arthropod communities in Ceuta are spatially structured according to fire occurrence. These differences are driven by species' functional traits [23], given that in burnt areas we found species more characteristic of open areas. By contrast, species more typically found in vegetated habitats, especially with shorter dispersal distances, were found in unburnt plots. According to our results, conservation managers should focus on individual taxa rather than total insect biodiversity [102], since our study's taxa demonstrated different overall responses to fire. Additionally, future research that examines the impact at multiple life stages when faced with different fire intensities, in different seasons, and over longer time scales will be essential to establish details pertaining to overall and specific arthropod responses to fire [22].

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/fire6050206/s1, Table S1: List and total abundance of Hymenoptera Formicidae, Coleoptera and Orthoptera species collected in pitfall traps from Ceuta (Spain).
Author Contributions: M.E.K., B.C., S.F. and X.S. conceived and designed the research; X.S. collected the field data; X.S. classified animals to order level; A.T., J.L.R. and P.B. classified animals to species level; B.C. analyzed the data; M.E.K., B.C. and X.S. participated in writing-original draft preparation, M.E.K., B.C., S.F., A.T., J.L.R., P.B. and X.S. participated in writing-review and editing. All authors have read and agreed to the published version of the manuscript.