Niche partitioning in tenebrionid species ( Coleoptera : Tenebrionidae ) inhabiting Mediterranean coastal dunes

We analyzed the abundance, distribution and niche overlap of species (Pianka’s Ojk index) in tenebrionid beetle communities inhabiting different biotopes in Tyrrhenian and Adriatic sand dunes. The rank abundance distribution of the different species has the form of a geometric series in both communities as predicted by the niche preemption hypothesis for communities in harsh environments. Mean niche overlap values did not deviate signifi cantly from null expectations, which indicates random interspecifi c interactions. These results, coupled with evidence of species habitat preferences, led us to conclude that the community organization of tenebrionid species inhabiting coastal dunes is determined more by habitat preferences than interspecifi c competition.


INTRODUCTION
One of the most controversial issues in insect ecology is the role of interspecifi c competition in determining community organization (Price et al., 2011;Schowalter, 2011).Research in this fi eld is usually diffi cult because of the large number of species of insects in a community and the many factors determining community structure.For these reasons, important insights into the role of niche partitioning in insect community structure may come from studying simple systems.
Thanks to the small number of species of insects and low environmental diversity in coastal dunes, these environments are excellent model systems for studying insect community organization.Coastal sand dunes are widely recognized as harsh environments, with high levels of stress and disturbance, and characterized by high salinity, drought, high temperatures, high wind speeds, low vegetation cover and low productivity (Martínez & Psuty, 2004;Feola et al., 2011), and hence host a low number of species of insects that are able to cope with these extreme conditions (McLachlan, 1991;Fattorini, 2008).
Among the beetles inhabiting Mediterranean sand dunes, tenebrionid beetles (Coleoptera: Tenebrionidae) are the most conspicuous component in terms of species richness, individual abundance and biomass (see Fattorini, 2008 for

Species-abundance distribution and spatial organization
We studied the overall species abundance distribution patterns at the two sites using a rank-abundance curve (May, 1975;Magurran, 1988;Hayek & Buzas, 2010).Since the geometric series is the reference model for communities composed of a few species and with high dominance (Fattorini, 2005), we fi rst used a χ 2 test to determine if our data deviated signifi cantly from the species abundance value predicted by this model.As our data did not deviate from the expected distribution, we used Ordinary Least Squares (OLS) regression described by Fattorini (2005) to model the rank-abundance curve.This approach is based on the fact that, if species are ranked from the most to the least abundant, and abundances are logarithmically transformed, a geometrical series exactly follows a strength line, which can be fi tted using an OLS regression.The coeffi cient of determination R 2 can then be used as a goodness-of-fi t measure.The geometric series is the mathematical model used to express the "niche preemption" hypothesis, in which the sizes of the niche hypervolumes (measured by species relative abundances) are sequentially preempted by the most abundant to the least abundant species.The fi rst species in the sequence occupies a fraction k of the resource hypervolume, the second species a fraction k of hypervolume not occupied by the fi rst, and so on.The niche preemption parameter k was calculated following He & Tang (2008).
We compared the slopes of the two regression lines using AN-COVA.We used the same approach to compare rank-abundance curves for the three biotopes at the Tyrrhenian site and those of the embryonic shifting dune biotope at the two sites.We refrained from other comparisons because of the small number of species occurring in the other biotopes at the Adriatic site.Calculations were done using PAST 3 (Hammer et al., 2001).
To determine if there is an association between species and habitat (i.e. if different proportions of the species occurred in different biotopes) we applied a χ 2 test to a species × biotope contingency table.To reveal the main spatial organization, variation in the species distributions at the two sites and in the different biotopes were analyzed using Detrended Correspondence Analysis (DCA) with CANOCO software (ver.4.5A) (Ter Braak & Šmilauer, 2002).

Niche overlap and species segregation
To express niche overlap between species pairs, we calculated the Pianka (1973) index of niche overlap using EcoSimR 1.00 (Gotelli & Ellison, 2013): where O jk is Pianka's index of niche overlap between species j and k, p ij is the proportion of the ith resource used by species j, p ik is the proportion of the ith resource used by species k, and n is the total number of resources.
This index is a modifi cation of the asymmetric index proposed by MacArthur & Levins (1967) for estimating competition coefficients from fi eld data on resource utilization.It is an analogue of a correlation coeffi cient and ranges from zero (complete dissimilarity in resource utilization between two species) to one (complete identity in resource utilization).Pianka's symmetric measure is more convenient than the original MacArthur and Levins' asymmetric form (May, 1974).Moreover, there is a general agreement that overlap measures cannot be used as true competition coeffi cients (Hurlbert, 1978;Abrams, 1980;Holt, 1987) and it is tion in these communities follow a simple niche preemption model; (2) if there is any non-random pattern in niche overlap, and (3) if species abundance distribution is related to inter-specifi c competition or other factors.

Study site and insect sampling
We studied community organization of tenebrionid beetles in two geographically separated dune systems in Central Italy: one on the Tyrrhenian coast (Montalto Marina, Latium region) and the other on the Adriatic coast (Campomarino, Molise region).Coastal zones are under severe human pressure, which may alter profoundly the structure of their insect communities (Fattorini, 2008).To minimize the possibility that the patterns are the result of anthropogenic infl uence we selected areas for study in relatively undisturbed coastal zones.As reported in previous ecological investigations, both of these sites have a relatively good conservation status (Acosta et al., 2009;Carboni et al., 2009).
At both sites, sampling was carried out along three transects separated by at least 100 m following a protocol similar to that of Fattorini et al. (2012).Along each transect, we identifi ed three biotopes as defi ned by the European Commission (1992Commission ( , 2007) ) (in parentheses EC codes): Embryonic shifting dunes (2110); shifting (white) dunes along the shoreline with Ammophila arenaria (2120); and Malcolmietalia dune grasslands (2230).These three biotopes were identifi ed in the fi eld based on the dominant and diagnostic plant species as indicated in the Habitat Directive Interpretation Manual (Santoro et al., 2012).In particular, the biotope corresponding to EC habitat 2110 is the closest to the seashore and is colonized by pioneer perennial plants such as Elymus farctus; the EC habitat 2120 occurs further inland and is the main dune ridge, which is mainly characterized by the presence of the dune-forming tussock grass, Ammophila arenaria; the EC habitat 2230 is even further inland and is in the transition zone between mobile and fi xed dunes, and is colonized by many annual grasses (Acosta et al., 2005(Acosta et al., , 2009;;Carboni et al., 2011).In each biotope, sampling was done using square plots of 2 × 2 m and sieving a standard volume of sand.We fi rst collected by hand all the beetles that were active on the ground in order to reduce the risk of their leaving the plot before being sampled.Hand sampling was done by systematically searching for beetles on the ground, under leaves and at the base of the plants.Then, we collected a fi xed volume of 12 l of sand.This volume was collected by dividing the 2 × 2 m plot into four units of 1 m 2 and collecting a sample of 3 l of sand from the centre of each unit.Sand was collected by digging a surface area of about 400 cm 2 to a depth of 2-3 cm.To minimize the effect of the sampling on the abundance of animals, large sized tenebrionids (Pimelia Fabricius, 1775, Erodius Fabricius, 1775, and Tentyria Latreille, 1802) were placed in tubes, identifi ed by an expert and released close to the plot where they were collected after all the plots at the same site were sampled.Small tenebrionids were identifi ed in the laboratory.At each site adults were sampled on one day in spring and one in summer and the data were pooled for each biotope because of the small numbers of species and individuals collect along each transect and in each plot.
Most species occurred in all three biotopes (see co-occurrence analysis below), although with different abundances, and the species recorded in the plots of the same biotope were very similar.In contrast, the environmental characteristics of the three biotopes were different.Thus, the biotope was assumed to be the most appropriate scale for this study and all analysis were conducted at this scale.
suggested that there is an inverse relationship between competition and niche overlap (Pianka, 1976(Pianka, , 1978)).For these reasons, MacArthur and Levins' original measure is now largely replaced by Pianka's symmetrical version.Indeed, niche overlap may refl ect intense competition for shared resources or, alternatively, a surplus of resources and the absence of competition (Glasser & Price, 1988).Thus, we do not consider values obtained from Pianka's index as "competition coeffi cients," but merely as measures of niche overlap (see Colwell & Futuyma, 1971;Pianka, 1974 for the distinction between overlap and competition).
To assess if the mean values of niche overlap recorded at each site were different from those expected by chance, we compared the observed values with the expected averages obtained from 10,000 simulated null-assemblages (pseudo-communities).This number of permutations is enough to avoid algorithm biases in calculations (Lehsten & Harmand, 2006).
Null-assemblages were simulated using Monte Carlo randomization algorithms that assign resource use values (in our case, number of individuals from different biotopes) to each species.The choice of an appropriate model to construct null-assemblages is a critical issue.Lawlor (1980) developed four randomization algorithms that differ in whether utilizations are reshuffl ed or replaced by a random number and whether the zeros in the matrix are retained or not.These algorithms are referred to as RA1, RA2, RA3 and RA4.Both retaining/relaxing niche breadths and retaining/reshuffl ing zeros have implications for the structure of the null community and affect the power of the test (Gotelli & Graves, 1996).Theoretical and empirical analyses of these algorithms (Winemiller & Pianka, 1990;Kobayashi, 1991) led to the conclusion that Lawlor's (1980) RA3 is probably the best algorithm for use in resource overlap null models.This algorithm tests for community structure by retaining niche breadth (i.e. the amount of specialization) for each species (simulated specialization equal to the observed value), but reshuffl es zero states (i.e. by randomly varying the particular resources that were used), thus destroying the guild structure manifested by the zero structure of the resource utilization matrix.
However, RA3 tends to overestimate niche overlap if the equiprobability assumption is not met, because more abundant resources will be used by all species even if niche segregation occurs.Thus, we also used, for comparative purposes, the RA2 algorithm, which tests for structure in the generalist-specialist nature of the resource utilization matrix by conserving guild structure (zero states are retained, thus preventing species that did not use a certain resource in the fi eld from doing so in simulations), but relaxes niche breadth (thus assuming a random equiprobable specialization) (Gotelli & Graves, 1996).Structure was assumed when P obs < exp = 0.05 or less (Gotelli & Graves, 1996).In all cases, equiprobable resource use was a-priori assumed in the analyses, because resource availability data were not available.To investigate between-site differences in niche overlap, we compared mean overlap values using Student's t-test.
We also calculated species segregation using Stone and Robert's (1990) C-score.The C-score for a species pair jk is calculated as: where R j is the row total for species j, R k is the row total for species k, and SS is the number of samples that contain both j and k.Thus, for any particular species pair, the C-score is a numerical index that ranges from a minimum of 0 (maximally aggregated) to a maximum of R j R k (maximally segregated with no shared samples).The matrix-wide C-score is an average of all the pairwise values of C-score for different species, so it refl ects both positively and negatively associated species pairs.To establish whether the matrix had an average C-score signifi cantly different from what can be expected from a null-model, we compared the observed values with the expected averages obtained from 10,000 simulated null-assemblages using the FF algorithm, which preserved row and column totals (Ulrich & Gotelli, 2007).For comparative purposes, we also used the fi xed row-equiprobable column (FE) algorithm, which preserved row totals (species occurrences) but treated columns (sampled biotopes) as equally probable, thus allowing species number per biotope to vary in the null assemblages.
High average C-scores indicate a lower randomness, i.e. a greater likelihood that the distribution of one species has been directly affected by the presence of other species.If a community were structured by competition, we would expect the C-score to be large relative to a randomly assembled community.While niche overlap analysis uses a data matrix with species-and sitespecifi c observed frequencies, the C-score searches for a nonrandom structure in the species assemblages by using a presence/ absence data matrix (Gotelli & Graves, 1996).

RESULTS
At both sites, the overall species abundance distribution did not deviate signifi cantly from that expected according to the geometric series model, with k = 0.487 for the Tyrrhenian site (χ 2 = 6.159, df = 6, P = 0.291) and k = 0.475 for the Adriatic site (χ 2 = 0.250, df = 6, P = 0.970).
Overall mean niche overlap at the Tyrrhenian site was 64% and 56% at the Adriatic site (Table 2).At both sites, the observed mean values were not signifi cantly lower than the respective simulated assemblages, regardless of the randomization algorithm used (Table 2).
Overall mean niche overlap values were not signifi cantly different at the two sites (t = 1.087, df = 40, P = 0.284).
An analysis of co-occurrences using the FF algorithm revealed that at both sites species distribution among biotopes did not deviate signifi cantly from a random pattern, with observed C-scores even lower than those expected [observed C-score = 0.143, mean of simulated C-score = Table 1.Abundance of tenebrionid species recorded in three biotopes along transects in Tyrrhenian and Adriatic dunes.A -embryonic shifting dunes; B -shifting dunes along the shoreline with Ammophila arenaria; C -Malcolmietalia dune grasslands.We found an association between species abundance and biotopes in both the Tyrrhenian (χ 2 = 46.071,df = 12, P = 0.001) and Adriatic (χ 2 = 32.932,df = 12, P < 0.0001) communities, which indicates that species have different habitat preferences.

Biotope
The lengths (in SD units) of the DCA gradients (Fig. 2) were as follows: 2.734, 0.735, 0.603 and 1.149.This indicates that the fi rst gradient is very important and thus the variation in the spatial distribution of species in tenebrionid assemblages is largely explained by the fi rst axis.Axis 1 explained 59.3% and Axis 2 7.1% of the variance.Axis 3 explained only 0.8% of the variance.The total explained variance was 67.4 and the sum of all egeievalues was 1.107.

DISCUSSION
Tenebrionids are the most abundant beetles in Italian beach-dune systems (Fattorini, 2008).The tenebrionid beetles living in Mediterranean coastal dunes include both omnivorous-opportunistic, large species, with a low population density (such as Pimelia bipunctata Fabricius, 1781, Tentyria grossa Besser, 1832, Erodius siculus Solier, 1834) and detritivorous, medium or small species with a high population density [such as Ammobius rufus (Lucas, 1846)].Obviously, not all small species are abundant.The small A. rufus is always very abundant, but T. aphodioides, another very small species, is a less abundant species (Fattorini, 2008).This is also the case for the communities investigated in the present study, in which A. rufus accounted for 21% of individuals at the Tyrrhenian site and 53% of those at the Adriatic site, respectively, whereas T. aphodioides accounted for 5% at the Tyrrhenian site and 1% at the Adriatic site, respectively.The low frequency of T. aphodioides, a burrowing but winged species, could be related to the windy nature of coastal environments.At both of the sites studied, we recorded a total of 7 tenebrionid species, a richness value similar to that recorded at other well preserved coastal sites in Italy (usually less than 10 species: Fattorini, 2008).
Tenebrionid communities in Mediterranean sand dunes are also characterized by a very simple structure, with a high dominance of certain species (Fattorini & Carpaneto, 2001;Carpaneto & Fattorini, 2001, 2003;Fattorini et al., 2012).Simple communities in early stages of successions, as well as those that are subject to strong pressures or are energetically poor, where r-selected species are favoured, are known to have species abundance distributions that follow the geometric series (Whittaker, 1965(Whittaker, , 1972;;Bazzaz, 1975;Gray, 1981;Giller, 1984;Nummelin, 1998;Keeley & Fotheringham, 2003).As already found for tenebrionid assemblages in other sand dune systems (see Fattorini, 2008), the geometric series provided an excellent fi t for both of the communities studied.The slopes of the two regression lines were also similar, which indicates that the two communities have an analogous pattern of species abundance distribution and the resource hyperspace is divided in a similar way in the two communities.This is confi rmed by a comparison of the embryonic shifting dune biotope, for which there are similar slopes for the two sites.In terms of the parameter k, which indicates the sequential, constant proportion of the total number of individuals in a community, it was about 0.5 for both systems, which indicates a rather sharp decline in the species niche hypervolumes, which is expected in communities where resources are divided according to the preemption theory.This decline was particularly strong in the Malcolmietalia dune grassland biotope at the Tyrrhenian site, because of the strong dominance of P. bipunctata, a very large, opportunistic beetle that also occurs in relatively highly anthropised areas.
The two tenebrionid communities studied not only had similar species abundance distribution patterns, but also similar values of mean niche overlap.This indicates that the two communities respond in a similar way to analogous constraints along the sea-inland environmental gradient, although located in very different geographical regions.
A community structure with P (observed < expected) = 0.05 or less is considered compatible with the occurrence of interspecifi c competition, whereas signifi cant aggregation of species, e.g.thanks to an unlimited resource, is indicated when P (observed > expected) = 0.05 or less (Gotelli & Graves, 1996).Based on this the two tenebrionid communities are not structured by interspecifi c competition because their mean niche overlap values were not sig- The role of competition in structuring tenebrionid communities is poorly explored.For example, different daily temporal strategies are evoked as a possible factor reducing competition between the dune species T. grossa and P. bipunctata (Fallaci et al., 1994) and a possible spatial separation has been suggested for two closely related syntopic congeneric species (see Fattorini, 2008).Of course, these forms of separation cannot operate in the species studied, because they were sampled simultaneously and no congeneric species were found at the same site.A study of a tenebrionid community in coastal sand dunes at El Saladar (Alicante, SE Spain) revealed substantial ecological overlap and no clear morphological evidence for interspecifi c competition, which indicates that competitive exclusion did not affect the species studied (Cantarino & Roman, 1991).
Our study was limited to one taxonomic assemblage, thus we cannot speculate about other species in other taxa that might compete for resources with tenebrionids.During sampling, however, we observed very few other macro-arthropods, such as spiders, larvae of myrmeleontids, anthicid beetles, curculionid beetles, scarab beetles of the tribe Psammodini and hemipterans.Spiders and larvae of myrmeleontids are predators, whereas curculionid beetles and hemipterans are phytophagous.Psammodini beetles and anthicid beetles are detritivores.However, Psammodini beetles were very rarely found and anthicid beetles are much smaller than tenebrionids, so competition, if it occurs, is likely to be very low.
Although a non-random structure of species assemblages does not necessarily imply competitive interactions, communities regulated by intense competition should exhibit non-random structures (Gotelli & Graves, 1996;Gotelli, 2000;Gotelli & McGill, 2006).This study revealed that mean niche overlap between species of tenebrionids in coastal dunes were not signifi cantly different from that expected by chance, which indicates that these communities are not regulated by competition.This is also supported by co-occurrence analysis, which indicates that there is no species segregation.The most obvious factor that is likely to infl uence the abundance of tenebrionid species in sand dunes is food availability.All the tenebrionid species studied are generalist detritivores that may feed on a variety of decaying matter with no particular specialization (Fattorini, 2008).This lack of trophic specialization is consistent with the high niche overlap recorded between species and lack of segregation.
On the other hand, as already recorded in previous studies (Fattorini, 2008;Fattorini et al., 2012), we also found that the abundance of tenebrionid species varied signifi cantly in the three biotopes studied [i.e.embryonic shifting dunes (EC habitat 2110), shifting dunes with Ammophila arenaria (EC habitat 2120) and Malcolmietalia dune grassland (EC habitat 2230)], which indicates that different species have different habitat preference.Because most species at the two sites sampled occurred in all three biotopes, spe-cies-biotope associations are expected to be weak.Thus, it is not surprising that the distribution of the species in the correspondence analysis plot is scattered.However, it is interesting to note that A. rufus is placed very close to the Malcolmietalia biotope at the Adriatic site (where it is the most abundant species) and P. bipunctata and H. pellucida to embryonic shifting dunes and Ammophila biotopes at the Tyrrhenian site.
The two sites were clearly separated by the correspondence analysis, based on their different species compositions, but the three biotopes, at each site, clustered in a similar way in the space defi ned by the two main axes, which further indicates that the two communities respond in a similar way to analogous constraints, although located in very different geographical regions.Thus, the comparative evidence of (1) the dominance of the most common species, highlighted by the geometric series distribution, (2) the absence of resource partitioning based on niche overlap and species segregation null models and (3) the differences in habitat preferences of the species revealed by the species abundance distributions, led to the conclusion that community organization in tenebrionid species inhabiting coastal dunes is determined by habitat preferences rather than by interspecifi c competition.

Fig. 1 .
Fig. 1.Rank-abundance plot of the tenebrionid beetles inhabiting Tyrrhenian and Adriatic beach-dune systems.Species are ranked from the most to the least abundant (x-axis).Species abundances are log-transformed (y-axis).Data were fi tted using ordinary least squares regressions.Figure 1a refers to overal rank-abundance curves for the Tyrrhenian (in black) and Adriatic (in grey) sites.Figure 1b presents rank-abundance curves for the three biotopes (A, B and C; see Table 1) at the Tyrrhenian site.Figure 1c compares rank-abundance curves for biotope A at the Tyrrhenian (in black) and Adriatic (in grey) sites.

Table 2 .
Pianka niche overlap indices recorded for tenebrionid communities in Tyrrhenian and Adriatic sand dunes.RA2 (relaxed niche breadth) and RA3 (retained niche breadth) are the two algorithms used to construct null-assemblages.