Niche partitioning and coexistence of two spiders of the genus Peucetia (Araneae, Oxyopidae) inhabiting Trichogoniopsis adenantha plants (Asterales, Asteraceae)

Niche theory suggests that the coexistence of ecologically similar species in the same site requires some form of resource partitioning that reduces or avoids interspecific competition. Here, from July 2013 to December 2015, we investigated spatial niche differentiation at three different scales of two sympatric congeneric spiders, Peucetia rubrolineata and P. flava, along an altitudinal gradient in shaded and open areas in an Atlantic forest in Serra do Japi, SP, Brazil. These spiders are peculiar in that they present an exclusive association with the plant Trichogoniopsis adenantha (Asteraceae). In theory, the coexistence of two Peucetia species could be explained by: (1) microhabitat segregation with individuals from different species occupying different parts of the same plants; (2) mesohabitat segregation with different species using plant in different environments; (3) macrohabitat segregation, where different species would not co-occur along the altitudinal gradient. With respect to micro-habitat use, in both species, different instars used different plant parts, while the same instars of both species used the same type of substrate. However, the two Peucetia species segregated by meso-habitat type, with P. rubrolineata preferring T. adenantha plants in shaded areas and P. flava preferring those in open areas. Our results support the hypothesis of niche partitioning begetting diversity, and highlight the importance of analyzing habitat use at multiple scales to understand mechanisms related to coexistence.


Introduction
The mechanisms involved in species coexistence are a central theme in ecology, as they are responsible for maintaining high species diversity in ecosystems worldwide [1][2][3][4][5]). Competitive interactions are stronger between morphologically similar and phylogenetically close sympatric species [6]. The first mathematical models of resource competition proposed that when a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 two species compete for the same resource, one species invariably eliminated the other [7][8][9]. However, considerable species diversity can be observed coexisting and persisting in the same trophic level of a given web of species interactions [10][11][12]. Coexistence theory proposes that the coexistence of ecologically similar species at the same site requires a form of resource partitioning [13].
One particular group whose coexistence mechanisms have been given considerable attention are spiders [21][22][23][24]. Spiders are one of the most diverse group of terrestrial predators [21,25], with up to 4 species and 131 individuals co-occurring per m 2 in tropical forests [26][27][28][29][30]. Partitioning of resources among spiders is generally explained by: (1) variation in prey types or sizes as a consequence of different hunting strategies or body size [31][32][33][34] reducing interspecific competition and allowing coexistence among species; (2) activity time, with different species varying in phenologies or circadian cycles [34][35][36] resulting in different population densities of the species throughout the year which reduces interspecific competition; (3) use of space, with different species using different microhabitats as the result of cryptic colorations [37], different foraging behaviors or different sites chosen for the placement of webs (e.g., different heights) [38][39][40][41][42], physiological tolerance [11,43] or based on thermal preference in a gradient of environmental characteristics in cave-dwelling spiders [44][45]. (4) Indirect influence of a species-specific fungus parasite, where the intensity of competition for space is reduced when the dominant host species is highly affected by the parasite and allows the colonization of other less dominant species [46]. (5) Predation, when agonistic interactions (e.g. predation by birds or intraguild predation) determine habitat occupation and niche assembly in tree-dwelling spiders [47]. In the literature, there is controversy over temporal segregation being one of the key mechanisms of coexistence in spider communities. On the one hand, [48] and [49] suggested that temporal segregation is an important factor that reduces niche overlap. [50] also argued that temporal segregation may facilitate the coexistence of species because the peak of abundance of each population will occur at different times, which reduces interspecific competition. For example, the spiders Meta menardi and Metellina merianae (Tetragnathidae), in addition to presenting distinct abundances throughout the year, also present distinct hunting strategies; M. merianae combines foraging outside and inside the web whereas M. menardi feeds exclusively on prey that fall into the web [34]. Temporal segregation may also occur when species differ in periods of activity during the day [51].
Despite the variety of mechanisms mediating coexistence in spiders, only a few studies have assessed different niche dimensions simultaneously in sympatric and ecologically similar species (e.g. [50,[52][53]). Understanding how different partitioning mechanisms acting through multiple niche dimensions contribute to the coexistence of ecologically similar organisms can help us advance in the understanding of the processes shaping diversity patterns in local scales. Here we study different aspects of habitat use of two co-occurring spiders in the genus Peucetia (Oxyopidae): Peucetia rubrolineata Keyserling, 1877 and P. flava Keyserling 1877. Peucetia is a cosmopolitan group, comprising 47 known species, with most species occurring in tropical regions [54][55]. Spiders of this genus do not build webs, weaving silk threads leading to the branches, flowers, or leaves of plants where they live. Some Peucetia species were observed showing strong associations with more than 50 species of plants belonging to 17 families, all with glandular trichomes, showing their strong associations with this plant characteristic [56]. Both species P. rubrolineata and P. flava occur from Venezuela to Argentina [55,57] and in the Serra do Japi-Brazil, P. rubrolineata and P. flava are found exclusively associated with the same plant: Trichogoniopsis adenantha (DC) (Asteraceae) [56,58]. The two spiders seem to use the same micro-habitat and food resources [57] and have similar phenologies and population fluctuation through the year [58]. Considering all ecological similarities, this pair of species living in the same plant comprises a promising study system to understand niche partitioning. In this sense the question we address here is: Is there niche partitioning at any level that helps explaining the coexistence of these two species? There are some possibilities that could explain the coexistence of these two Peucetia species: (1) microhabitat segregation-whereby P. rubrolineata and P. flava would occupy different parts of the plant (e.g., leaves, stems, and flower heads) avoiding competition and allowing coexistence even while inhabiting the same plant. 2) mesohabitat segregation-in which Peucetia species exhibit differential distribution based on luminosity and humidity where the host plant is found. 3) macrohabitat segregation-P. rubrolineata and P. flava may be distributed differently along an altitudinal gradient. On the one hand, in the field these two species apparently occur in the same specific parts in the host plant (microhabitat), occur in similar proportions along the altitudinal gradient (macrohabitat) and their dynamics and phenologies are similar [58]. On the other hand, these two species of spiders are observed in places with different light levels even if they occupy the same plant, P rubrolineata occurring more in shaded places and P. flava occurring more in open places. Our hypothesis is that there is segregation at a different spatial level (mesohabitat), where one species occurs more often in shaded places and the other species occurs more often in open places due to its physiological tolerances. However, the other possibilities of habitat segregation (micro and macrohabitat) will also be tested in the present study.

Study area
This study was carried out at Serra do Japi, located between 23˚11' S and 46˚52' W of the Atlantic Plateau between the municipalities of Jundiaí, Itupeva, Cabreúva, Pirapora do Bom Jesus, and Cajamar in the state of São Paulo, Brazil. Located between 700 and 1,300m above the sea level, this area (354 km 2 ) comprises seasonal mesophyllous forests. The climate is seasonally well defined, with average monthly temperatures varying from 13.5˚C in July to 20.3˚C in January and with a rainy season in summer (December-March) and dry season in winter (June-August) [59]. Surveys were conducted in different regions of the Serra: (1) at the Department of Water and Sewage (DAE) dam; (2) near the field base at the area locally known as "Biquinha"; (3) along a regional pathway at a locality named "TV cultura" at two altitudes (900-1000 m and 1170-1290 m). To determine the role of (1) microhabitat-stratification on plants, and (2) mesohabitat-patches with different environmental conditions in resource partitioning, studies were conducted from 2013-2015. To determine the role of macrohabitataltitudinal distribution, research was carried out in March 2014 and March 2017. At the end of the Materials and Methods, there is a table summarizing all methods used, dates, and specific record locations ( Table 1).

Occurrence of Peucetia rubrolineata and P. flava in Trichogoniopsis adenantha
During the period between July 2013 to December 2015, we inspected 300 T. adenantha in open and shaded areas (S1 Table). For each plant with spiders, we registered the species of spider, the instar of development and the specific part in the plant where the spider was found. In each monthly survey the 300 plants were randomly selected so that these 300 plants could be different each month. The two species of Peucetia can be easily recognized in the field by the different pattern of staining on the cephalothorax and abdomen. In P. rubrolineata a continuous double line can be observed in the cephalothorax and in P. flava that same line is cracked (Fig 1). We identified the instars in accordance with the guidelines published by [60]. This classification was based on the size of the cephalothorax and the first pair of legs (For details see [58,61]. We collected some individuals from both species of Peucetia for later corroboration of identification with the specialist (Brescovit A.D.),

Microhabitat-Stratification on plant
Phenology and availability of types of branches of the plant Trichogoniopsis adenantha. To uncover the phenology of T. adenantha, of the 300 plants previously recorded we selected 20 plants randomly in each month and registered the number of vegetative branches and the state of development of flower heads. Once the phenology of the plant was recorded, we could evaluate if the phenology affected the occurrence of instars of P. rubrolineata and P. flava on different branches of the host plant.
We classified flower heads in situ into five phenophases as described by [62] and [63]: (F1) closed bud, very small, bracts covering the whole bud; ( Table). Subsequently, based on the predominance of the phenophases of each flower head, we classified the branches into the following categories: (1) vegetative branches; (2) branches with flower buds, with the mean number of flowers in the F1 or F2 phase greater than 50% of the total number of flower heads (owing to inter-individual variation in the number of flower heads from 6 to 18 flower heads per branch); (3) branches with open flowers, with the mean number of flowers in the F3 greater than 50% of the total number of flower heads; (4) type 4 branches, with the mean number of flowers in the F4 greater than 50% of the total number of flower heads; (5) type 5 branches, with the mean number of flowers in the F5 greater than 50% of the total number of flower heads.
Distribution of Peucetia rubrolineata and Peucetia flava instars on Trichogoniopsis adenantha branches. To verify whether the distribution of the P. rubrolineata instars was similar in the different types of branches, we compared the frequency of the types of branches available with the branches actually occupied by the instars of the spiders. For this analysis, we used only the information from the months where P. rubrolineata was most abundant within the two-and-a-half-year period previously recorded. The same procedure was performed for P. flava.
Micro-site on the branch of the Trichogoniopsis adenantha plant. To verify whether instars of P. rubrolineata have the same pattern of distribution in different specific parts of the plant (leaf, stem, and flower head and dry flower head), we used the Chi-square Test. Here again, we used the data from the same two-and-a-half-year period previously recorded. As the null hypothesis, we predicted the spiders of each instar would occur on similar parts of the branch with similar probability. We also adopted the same protocol for P. flava.
Similarity in the distribution of instars of the two Peucetia species on Trichogoniopsis adenantha. We also investigated whether the same instars of both species occupy the same part of the plant and used the Chi-squared Test to examine the significance of observed differences in the frequency of occupation. To do this, we related the total available parts of the inspected plants to the parts that were actually occupied by the same instar of both Peucetia species.

Mesohabitat-Patches with different environmental conditions
Plant habitat. In a separate survey, we recorded T. adenantha plants that had at least one individual of some of the two species of Peucetia until reaching a total of 100 records of individuals of each species. In those plants in which an individual of Peucetia was registered we set a camera with a fisheye lens to photograph the cover vegetation above each plant. Subsequently, we divided photographs into sectors and classified each site according to canopy cover percentage classes: 0-20%; 21-40%; 41-60%; 61-80%, and 81-100%. We then tested if the distribution pattern of the two species of Peucetia differed among classes of vegetal cover, using a Chi-squared Test.
Co-occurrence of both spider species. To determine the degree of niche overlap of the two species of Peucetia in each instar, we used the Jaccard overlap index [64]: Where "A" is the number of plants where both species of spiders were present simultaneously, "B" is the number of plants where only P. rubrolineata was present and "C" the number of plants where only P. flava was present. To calculate the index, we pooled together all the surveyed plants where at least one individual of some of the spider species was present during the two-and-a-half-year period. The index was calculated for the same instar of the two species of Peucetia separately because the two species presented in similar proportions of abundance throughout the year and the same instars of the two species occupy the same type of branch on the plant. For all comparisons regarding the frequencies of use of habitat-temporal segregation, ontogenetic variation, microhabitat segregation and macrohabitat partitioning-we used a Chi-square test with Monte Carlo permutations with 10000 simulations [65] to test for significance considering an alpha value of 0.05. In cases of multiple pairwise comparisons we used the Williams´correction. For all statistical analyzes, we used the free software R [66]. The package "rmngb" was used for all Chi-squared analyses.

Phenology of Trichogoniopsis adenantha
In general, T. adenantha had both vegetative and reproductive branches throughout the year; however, the proportions varied with the season (Fig 2). In winter and spring, vegetative branches occurred in larger proportions. At the end of the dry season, usually in spring (months), there was growth of vegetative branches, followed by an increase in the number of reproductive branches, with a large number of buds in spring and summer. During March and April, there was a period with greater abundance of branches with open flowers. In May, a large number of flowering branches occurred in phase 4 (seed formation), and in June and July, there were several flowering branches in phases 4 and 5 dry (flower heads). In general, vegetative branches were more abundant than reproductive branches. Production of flower heads occurred throughout the year; however, the proportion of each phenophase varied from year to year (Fig 2).

Microhabitat-Stratification on the plant
Distribution of the Peucetia species on T. adenantha branches types. Young spiders of both species were more frequent on the vegetative branches than in reproductive ones (Chisquared test = P. rubrolineata: χ 2 = 9.37, P = 0.034; P. flava: χ 2 = 10.777, P = 0.005; in 10000 simulations). Juvenile spiders were more frequent on vegetative branches and on type five branch (Chi-squared test = P. rubrolineata: X 2 = 8.31, P = 0.043; P. flava: X 2 = 11.111, P = 0.009), as well as subadults, (Chi-squared test = P. rubrolineata: X 2 = 11.327, p = 0.005; P. flava: X 2 = 8.512, p = 0.006) although the other types of branches are more frequent. Adult individuals were more frequent on type 5 branches, especially in the inflorescences where shelters made with leaves and dry flower heads are often observed (Chi-squared test = P. rubrolineata: X 2 = 10.096, P = 0.0038; P. flava: X 2 = 12.791, P = < 0.001). Finally, spiderlings of the two Peucetia species were the only ones that presented similar observed and expected frequencies in all types of branches (Chi-squared test = P. rubrolineata: X 2 = 4.22, P = 0.215; P. flava: X 2 = 1.22, P = 1) (Fig 3). Regardless of the proportion of the different branches, the pattern of distribution of the instars of the two species of Peucetia in these branches remained the same throughout the year.  (Table 2). Second and third instars of P. rubrolineata were more frequently recorded on leaves, whereas fourth-instar spiders were observed on the leaves, stem, and flower heads, the fifth, sixth, and seventh instars occurred more frequently in the dry flower heads. Adults (eighth-instar) occurred more frequently in dry flower heads and in shelters made from remnants of the dry flower heads (Table 2; Fig 4).
Despite the difference in the occupation of the types of branches and parts of the plant among instars of the same species, when the same instars of the two species are compared, the pattern of occupation on the plant is similar (Table 3; Fig 5). Only spiderlings presented a different occupation pattern between species; P. rubrolineata spiderlings used preferentially the leaves, whereas P. flava spiderlings were mostly found in the stem and flower heads (Table 3; Fig 5).

Mesohabitat-Patches with different environmental conditions
Co-occurrence of both Peucetia species. The frequencies of the two species of Peucetia were different in the two types of vegetation cover (X 2 = 75.444, P = < 0.001; in 10000 simulations) (Fig 6). Peucetia rubrolineata was recorded at a higher frequency in T. adenantha plants located in environments with denser canopy (X 2 = 41.622, P = < 0.001) (Fig 6B). Conversely, P. flava was recorded more frequently in T. adenantha plants in more open environments (X 2 = 13.085; P = 0.01) (Fig 6A). In the intermediary levels of vegetation cover, the frequencies of both Peucetia species were similar (Fig 6).
In environments with intermediate canopy cover, a larger number of plants were observed with the two Peucetia species. Of the 729 plants that had spiders during the study period, 43.5% contained only P. flava and 53.3% contained only P. rubrolineata, whereas only 3.4% of the plants contained both species. The values of niche overlap varied from 0.024 to 0.059, with a mean of 0.034, indicating low overlap between the same instar of the two species of Peucetia (Table 4).

Macrohabitat-Altitudinal separation
Altitudinal distribution. Both Peucetia species were found at all altitudes at the study site, but in different frequencies (63 individuals of P. rubrolineata and 37 individuals of P. flava in total). However, these frequencies varied depending on availability of T. adenantha plants in more open areas (low canopy cover) or in more closed places (high canopy cover) and not with respect to the altitude. Sites with low canopy cover had mostly individuals of P. flava, whereas sites with high canopy cover had individuals of P. rubrolineata (Table 5). In the observation made in 2014 at site (2b), called the TV Cultura Pathway Region, T. adenantha plants were found with little canopy cover and a higher abundance of P. flava was recorded. Three years later, in 2017, a higher abundance of P. rubrolineata was observed (Table 5) in the same region and at the same altitude (2b � ), when vegetation grew and canopy cover increased on T. adenantha plants.

Discussion
Niche theory suggests that the coexistence of ecologically similar species requires some form of resource partitioning that reduces or prevents interspecific competition [13]. In the studied system, the two Peucetia species inhabiting plants of T. adenantha present similar phenology  and population dynamics [58]. In addition, the same instars of the two spider species had similar distributions across the different parts of the plant, indicating that the microhabitat use overlaps between the two species. However, our hypothesis that the two Peucetia species inhabiting T. adenantha plants are segregated has been confirmed, because they were distributed differently at the mesohabitat level. Peucetia rubrolineata occurred more frequently in places where canopy cover was high, whereas P. flava occurred more frequently in sites with open canopy. Although several studies have shown that interspecific competition is one of the main factors frequently observed in phylogenetically close spiders (e.g. [41]), in our study there is low overlap at the mesohabitat scale. Thus, we can infer from our results that competition may not be an important factor currently shaping these populations, although competition in the past might have favored habitat segregation patterns. Further research on sites where these species do not occur in sympatry could shed light on whether this segregation is a product of interspecific interactions or simply reflect different responses to abiotic conditions. Results also indicated that these species have different altitudinal distributions because P. rubrolineata was more abundant at lower altitudes in the Serra do Japi, whereas P. flava occurred more frequently at higher altitudes. However, this difference seems to be more related to the availability of plants in different environments (shaded and exposed) than related to the altitude. At higher altitudes, the frequency of plants in the sunny areas was higher, as was the frequency of P. flava. In contrast, at lower altitudes, shaded areas were more frequent and the frequency of P. rubrolineata was higher. At intermediate altitudes, the abundance of spiders depended on the abundance of plants in the sun and shade. At the same site, the abundance of P. flava and P. rubrolineata varied in different years depending on the degree of shading. Although the two Peucetia species occurred in environments with intermediate canopy cover, the co-occurrence of these spiders in the same host plant was infrequent. Although we observed a clear distribution pattern of the two species of Peucetia more related to the type of environment (open or shaded) than to altitude, this statement should be taken with caution, because the recorded abundances of the two species for each altitude were low. Future studies Niche partitioning in two sympatric Peucetia spiders would be necessary where the number of spiders collected by altitude was higher, and the low influence of altitude on the coexistence of the two species of Peucetia was corroborated. Although P. rubrolineata and P. flava are sympatric and their population dynamics are similar throughout the year, the general pattern of spatial segregation registered occurred in the two years of study. This type of segregation as a central mechanism in the coexistence of sympatric spiders has been recorded in other cursorial spiders of the genus Syspira (Miturgidae), where two species of Miturgidae also exhibited different spatial distribution related to differences in temperature and humidity at the site [67]. In a central Amazon forest, [50] recorded significant differences in the soil type preference (sandy and clay soils) of four species of Ctenus (Ctenidae). In Lycosidae, horizontal stratification appears to be a common mechanism of niche differentiation [68]. In the temperate region the relative abundance of Pardosa alacris gradually increased with the canopy opening (mesohabitat level), whereas the opposite trend was observed for P. lugubris. Thus, niche differentiation along the canopy opening gradient seems to mediate the coexistence of these two species in meta-communities [52]. The Lycosids Geolycosa xera archboldi and G. hubbelli also exhibited preferences for different types of vegetation (mesohabitat level) and for different percentage of litter cover (micro habitat level) [69]. These examples show that habitat segregation at the mesohabitat level, as we found here, may be an important mechanism maintaining the high diversity of arthropods, especially spiders, in forested habitats.
Spiderlings of the two Peucetia species were more frequent on the vegetative branches of T. adenantha. A number of predatory arthropods (e.g., others Peucetia spider species, the spider Misumenops argenteus, Thomisidae) are specifically associated with glandular trichome-bearing plants like T. adenantha, where they capture prey attached to the sticky thricomes [57,70]. Spiderlings are probably more frequent in plant parts with more trichomes, which would increase the number of prey trapped [71]. Moreover, trichomes provide some protection to the spiderlings from their enemies, such as other spiders [72]. Fourth-instar young spiders were observed preying on endophytic herbivores of the flower head of T. adenantha, especially the fruit flies Trupanea sp. (Tephritidae), whereas juveniles (fifth and sixth instars) consumed inchworms of Geometridae that feed on flower heads and other insects that inhabit the plant (see [60]). In the case of subadults and adult spiders, they occurred more frequently in the flower head of the plant, especially those branches that also have dry structures where they rest. This higher frequency likely occurred because, besides having a place for camouflaging, it is also easier for them to capture larger prey, such as floral visitors, which visit flower heads with open flowers at phase 3 and phase 4. In fact, [71] recorded adult spiders feeding on Pseudoscada erruca, Aeria olena, and Episcada carcinia (Ithomiinae), and all floral visitors on the flowers of T. adenantha. Additionally, adult females build the nest to place the egg sac using the plant's dried structures (personal observation). Thus, preferences for types of branches could be partially explained by the types of prey available and used as food by different instars as well as reproductive behavior. This ontogenetic habitat segregation found in both of the studied species may also decrease intraspecific competition among different instars of the same species, which would be reflected in the success of individuals in different stages of development.
Finally, because these two species of Peucetia are associated exclusively with T. adenantha, we expect that in the other parts of the Serra where we did not register these species, the same pattern of distribution and the same degree of spatial overlap occur in the different scales that we study. However, these species of Peucetia may be associated with other plants that have glandular trichomes, so if there are other plants in other places with glandular trichomes there would probably be other distribution patterns. However, further studies would be needed to compare the degree of effect of the availability of other plants on the distribution of Peucetia.
Examining the niche partition between two morphologically similar, sympatric spiders, with overlapping temporal distributions we show inter-and intraspecific segregation in habitat use at different spatial scales. Although P. flava and P. rubrolineata are occurring in sympatry with similar phenologies, at similar altitudes and using the same parts of the same plants species, the distribution of the T. adenantha plant in shaded and open environments affects the distribution of the two species corroborating our hypothesis. Our results support the hypothesis of niche partitioning begetting diversity, and highlight the importance of analysing habitat use at multiple scales to understand mechanisms related to coexistence.
Supporting information S1 Table. Phenology and abundance Peucetia species. This data set is about the phenology and abundance of the two species of Peucetia over the two and a half years. Each month we register 300 plants of Trichogoniopsis adenantha randomly to register the spiders. We recorded for each spider the developmental stage, sex and part of the plant where the individual was. (TXT) S2 Table. Phenology Trichogoniopsis adenantha plants. This set of data is on the phenology of T. adenantha plant. Of the 300 plants recorded, we randomly selected 20 plants each month to record the penology. In each registered plant the number of vegetative and reproductive branches was noted. Moreover, in each reproductive branch we note the phenophase of the flower heads. (TXT)