ECOLOGICAL SEGREGATION IN A SMALL MAMMAL HYBRID ZONE: HABITAT‐SPECIFIC MATING OPPORTUNITIES AND SELECTION AGAINST HYBRIDS RESTRICT GENE FLOW ON A FINE SPATIAL SCALE

The degree to which closely related species interbreed is determined by a complex interaction of ecological, behavioral, and genetic factors. We examine the degree of interbreeding between two woodrat species, Neotoma bryanti and N. lepida, at a sharp ecological transition. We identify the ecological association of each genotypic class, assess the opportunity for mating between these groups, and test whether they have similar patterns of year‐to‐year persistence on our study site. We find that 13% of individuals have a hybrid signature but that the two parental populations and backcrosses are highly segregated by habitat type and use. Also, we find that adult hybrids are comparable to parental types in terms of year‐to‐year persistence on our site but that, among juveniles, significantly fewer hybrids reach adulthood on site compared to their purebred counterparts. Our analyses show that this hybrid zone is maintained by occasional nonassortative mating coupled with hybrid fertility, but that these factors are balanced by lower apparent survival of juvenile hybrids and habitat‐based preference or selection that limits heterospecific mating while promoting backcrossing to habitat‐specific genotypes. This system presents a novel example of the role that sharp resource gradients play in reproductive isolation and the potential for genetic introgression.     

How isolated are Pleistocene refugia? Results from a study on a relict woodrat population from the Mojave Desert,California

Pleistocene vicariance is often invoked to explain the disjunct populations of animals in habitat refugia throughout the southwestern United States. The combined effects of small population size and isolation from the rest of the contiguous range are thought to result in genetic differentiation of relict organisms. Here, we describe a relict population of dusky‐footed woodrats (Neotoma fuscipes Baird) found in a pinyon‐juniper‐oak community in a small mountain range within the Mojave Desert. We compare morphological and genetic data for these individuals with two populations within the contiguous range, and with another species of woodrat (Neotoma lepida). We also examine the distributional overlap between contemporary oak species and dusky‐footed woodrats, and estimate the potential oak woodland habitat available during the late Quaternary. As expected, both the morphological and genetic analysis confirm that the relict population is N. fuscipes. Within the limitations of our data, we detect no evidence of differentiation. Instead, the relict population forms a paraphyletic group with the nearest population within the contiguous range. This may be explained by the combined influences of a shorter period of isolation and a greater effective population size than was originally expected. The linkage between contemporary oak and dusky‐footed woodrat distributions is very tight, reinforcing the idea of an obligate relationship between the two. We estimate that at ~8000 ybp, pinyon‐juniper‐oak woodlands may have covered ~53% of the central Mojave, forming large contiguous areas of habitat. Although considerably more fragmented, at present ~12% of the area consists of relict woodlands. Our results suggest that there may be numerous other woodrat refugia, with a relatively high degree of connectiveness between the larger ones. Animals within them may effectively function as a single metapopulation, buffering against occasional stochastic extinction events.     

Characterization of 18 microsatellite loci for woodrats of the Neotoma lepida group (Rodentia,Cricetidae, Neotominae)

The desert woodrat complex (Neotoma lepida group) is a set of four closely related species occupying the arid lands of western North America. We have developed 18 polymorphic microsatellite loci to analyse the population structure, mating patterns, and hybridization at points of contact between the contiguously allopatric species in the complex.     

Evaluating nest supplementation as a recovery strategy for the endangered rodents of the Florida Keys

The Key Largo woodrat (Neotoma floridana smalli) and Key Largo cotton mouse (Peromyscus gossypinus allapaticola) are federally endangered subspecies endemic to the tropical hardwood hammocks of Key Largo, Florida. Woodrats are considered generalists in habitat and diet, yet a steady decline in natural stick nests and capture rates over the past several decades suggests that they are limited by the availability of nesting habitat due to habitat loss and fragmentation. The more specialized Key Largo cotton mouse appears to rely on old growth hammock, a habitat type that is rare following past land clearing. In 2004, the U.S. Fish and Wildlife Service started building supplemental nest structures to restore habitat quality and connectivity for these endangered rodents, but nest use requires evaluation. We used camera traps and occupancy models to evaluate the factors influencing woodrat and cotton mouse use of the supplemental nests. We detected woodrats at 65 and cotton mice at 175 of 284 sampled nest structures, with co‐occurrence at 38 nests. Woodrat nest use followed a gradient from low nest use in the north to high nest use in the south, which might relate to the proximity of free‐ranging domestic cat (Felis catus) colonies in residential developments. Cotton mouse nest use, however, was related positively to mature hammock and related negatively to disturbed areas (e.g. scarified lands). The two species occurred independently of each other. Stick‐stacking behavior was observed at supplemental nests and, although it was correlated with detection of woodrats, it was not a strong predictor of their occurrence. We suggest that nest supplementation can be an important tool for species recovery as habitat quality continues to improve with succession.     

Experience matters: prior exposure to plant toxins enhances diversity of gut microbes in herbivores

For decades, ecologists have hypothesised that exposure to plant secondary compounds (PSCs) modifies herbivore‐associated microbial community composition. This notion has not been critically evaluated in wild mammalian herbivores on evolutionary timescales. We investigated responses of the microbial communities of two woodrat species (Neotoma bryanti and N. lepida). For each species, we compared experienced populations that independently converged to feed on the same toxic plant (creosote bush, Larrea tridentata) to naïve populations with no exposure to creosote toxins. The addition of dietary PSCs significantly altered gut microbial community structure, and the response was dependent on previous experience. Microbial diversity and relative abundances of several dominant phyla increased in experienced woodrats in response to PSCs; however, opposite effects were observed in naïve woodrats. These differential responses were convergent in experienced populations of both species. We hypothesise that adaptation of the foregut microbiota to creosote PSCs in experienced woodrats drives this differential response.     

Canopy insect herbivores in the Azorean Laurisilva forests: key host plant species in a highly generalist insect community

This article explores patterns of insect herbivore distribution in the canopy of the Laurisilva forests on seven islands in the Azores archipelago. To our knowledge, this is one of the first extensive study of this type in tree or shrub canopies of oceanic island ecosystems. One of the most frequently debated characteristics of such ecosystems is the likely prevalence of vague, ill‐defined niches due to taxonomic disharmony, which may have implications for insect‐plant interactions. For instance, an increase in ecological opportunities for generalist species is expected due to the lack of predator groups and reduced selection for chemical defence in host plants. The following two questions were addressed: 1) Are specialists species rare, and insect herbivore species randomly distributed among host plant species in the Azores? 2) Are the variances in insect herbivore species composition, frequency and richness explained by host plants or by regional island effects? We expect a proportional distribution of herbivore species between host plants, influenced by host frequency and distinct island effects; otherwise, deviation from expectation might suggest habitat preference for specific host tree crowns. Canopy beating tray samples were performed on seven islands, comprising 50 transects with 1 to 3 plant species each (10 replicates per species), giving 1320 samples from ten host species trees or shrubs in total. From a total of 129 insect herbivore species, a greater number of herbivore species was found on Juniperus brevifolia (s=65) and Erica azorica (s=53). However, the number of herbivore species per individual tree crown was higher for E. azorica than for any other host, on all islands, despite the fact that it was only the fourth more abundant plant. In addition, higher insect species richness and greater insect abundance were found on the trees of Santa Maria Island, the oldest in the archipelago. Insect species composition was strongly influenced by the presence of E. azorica, which was the only host plant with a characteristic fauna across the archipelago, whereas the fauna of other plant crowns was grouped by islands. The great insect occurrence on E. azorica reflects strong habitat fidelity, but only four species were clearly specialists. Our findings indicate a broadly generalist fauna. The simplicity of Azorean Laurisilva contributed to the understanding of insect‐plant mechanisms in canopy forest habitats.     

Molecular phylogenetics of an endangered species: the Tamaulipan woodrat (<Emphasis Type="Italic">Neotoma angustapalata</Emphasis>)

Neotoma angustapalata (Tamaulipan woodrat) is a large cricetid rodent found only in southwestern Tamaulipas and northeastern San Luis Potosí, Mexico. This species currently is listed as endangered due to habitat alteration, its restricted distribution, and relative rarity. Previous taxonomic assessments have allied N. angustapalata with N. albigula (now encompassing N. leucodon), N. mexicana or N. micropus. We sequenced portions of the mitochondrial cytochrome b gene from two skin samples of the Tamaulipan woodrat, including one of two topotypes. We estimated genealogical relationships between N. angustapalata and other species of Neotoma using maximum likelihood and Bayesian inference. In general, our results confirm the phylogeny of woodrats as proposed previously but we also recovered major genetic differentiation within what currently is recognized as N. mexicana and N. albigula. Our data document that the Tamaulipan woodrat is genetically indistinguishable from geographically adjacent haplotypes of N. leucodon. However, mitochondrial introgression from N. leucodon cannot be ruled out inasmuch as we were not able to obtain nuclear sequence data for N. angustapalata. Morphological analyses document that both male and female Tamaulipan woodrats differ morphologically from N. leucodon. Given that the Tamaulipan woodrat is diagnosable morphologically and occurs in habitat that differs from N. leucodon, we recognize N. angustapalata as a species-level entity.     

Evaluation of Factors Restricting Distribution of the Endangered Key Largo Woodrat

ABSTRACT We examined a suite of models in an information theoretic framework to identify factors restricting presence of the endangered Key Largo woodrat (Neotoma floridana smalli) throughout its remaining habitat. Models containing variables related to availability of nest sites and mammalian predator abundances were supported by our data. Abundance of natural (large overstory trees) and artificial (rock and debris piles) nest substrate were the most important predictor variables, followed by indices of feral cat and raccoon (Procyon lotor) abundance. We recommend increasing abundance of nest substrate in the short term through addition of artificial nest substrate and in the long term through continued protection of remaining forest habitat.     

Differences in dietary composition and preference maintained despite gene flow across a woodrat hybrid zone

1. Ecotones, characterized by adjacent yet distinct biotic communities, provide natural laboratories in which to investigate how environmental selection influences the ecology and evolution of organisms. For wild herbivores, differential plant availability across sharp ecotones may be an important source of dietary‐based selection.
2. We studied small herbivore diet composition across a sharp ecotone where two species of woodrat, Neotoma bryanti and N. lepida, come into secondary contact with one another and hybridize. We quantified woodrat dietary preference through trnL metabarcoding of field‐collected fecal pellets and experimental choice trials. Despite gene flow, parental N. bryanti and N. lepida maintain distinct diets across this fine spatial scale, and across temporal scales that span both wet and dry conditions.
3. Neotoma bryanti maintained a more diverse diet, with Frangula californica (California coffeeberry) making up a large portion of its diet. Neotoma lepida maintains a less diverse diet, with Prunus fasciculata (desert almond) comprising more than half of its diet. Both F. californica and P. fasciculata are known to produce potentially toxic plant secondary compounds (PSCs), which should deter herbivory, yet these plants have relatively high nutritional value as measured by crude protein content.
4. Neotoma bryanti and N. lepida consumed F. californica and P. fasciculata, respectively, in greater abundance than these plants are available on the landscape—indicating dietary selection. Finally, experimental preference trials revealed that N. bryanti exhibited a preference for F. californica, while N. lepida exhibited a relatively stronger preference for P. fasciculata. We find that N. bryanti exhibit a generalist herbivore strategy relative to N. lepida, which exhibit a more specialized feeding strategy in this study system.
5. Our results suggest that woodrats respond to fine‐scale environmental differences in plant availability that may require different metabolic strategies in order to balance nutrient acquisition while minimizing exposure to potentially toxic PSCs.

     

Implications of Woodrats and Other Scavengers for Avian Telemetry Studies

Abstract: During 2 years of radiotelemetry research on chukars (Alectoris chukar) in western Utah, USA, we found 28% of retrieved radios (n = 78) in rubbish nests of woodrats (Neotoma spp.). Such movement and disturbance of carcasses and radios by woodrats and other species has implications for radiotelemetry studies. We evaluated spatio-temporal movement of 51 radiocollars attached to chukar carcasses in western Utah. Most (80%) carcasses were scavenged within one week and by the end of 3 weeks 25 (50%) had been retrieved from woodrat middens. Scavenging activity can both obscure important clues needed to identify causes of mortality and bias telemetry studies by delaying onset of mortality signals.     

Differential regional response of the bushy‐tailed woodrat (Neotoma cinerea) to late Quaternary climate change

Aim To reconstruct the regional biogeographical history of the bushy‐tailed woodrat, Neotoma cinerea (Rodentia: Cricetidae), across its distribution using multiple sources of information, including genetic data, ecological niche models and the palaeorecord. Location Western North America. Methods We analysed complete cytochrome b gene (1143 bp) sequences from 182 specimens of N. cinerea using Bayesian and coalescent methods to infer phylogenetic relationships, time of major divergences, and recent demographic trends. For comparison, we developed clade‐specific ecological niche models for groups of interest and analysed spatial trends of N. cinerea in the palaeorecord as well as temporal frequency trends across strata of individual palaeomiddens. Results We found two largely allopatric clades within N. cinerea, with several regionally distinct subclades showing contrasting recent population dynamics. Niche models showed consistent habitat at the Last Glacial Maximum (LGM) and modern times in the Rocky Mountains and northern United States, while the Great Basin may have been markedly less suitable at the LGM than today. The palaeorecord showed great spatio‐temporal variability in the presence of N. cinerea, but documents broad‐scale patterns of occupancy and regional population trends. Main conclusions The Quaternary dynamics and evolutionary history of N. cinerea appear to have been shaped by both vicariant events associated with geographical barriers and the availability of suitable habitat through time. Divergence of the two major clades dates to the Pliocene–Pleistocene transition, with clades separated by the Green and Colorado rivers and northern Rocky Mountain Pleistocene glaciations. We found largely concordant genetic, niche model and palaeorecord patterns suggesting long‐term population stability in the Rocky Mountains, while extant clades in the Great Basin and far north appear to have expanded or re‐expanded into these regions relatively recently. Furthermore, disjunct haplotype distributions, regional demographic history, and historical distribution of suitable habitat suggest that the Great Basin has been a particularly dynamic region.     

Ecological and biodiversity gradients across alpine dry grassland habitats: implications for an endangered species

Dry grasslands are of great interest for nature conservation in Europe, because they have a central role in the conservation of numerous rare and endangered species. In this study carried out in the Brenta mountain group (Italian alps), we investigated the effect of environmental factors mainly controlled by topography, on the biodiversity trends across different dry grassland habitats where the threatened alpine stenoendemic Erysimum aurantiacum grows. Plant community data and ecological factors were analysed by means of a multi‐habitat CCA approach and by analysis of biodiversity gradients in 7 natural and semi‐natural habitats. We found that species turnover and biodiversity patterns vary as a function of multi‐factorial ecological gradients. For the single habitats, elevation gradient was the main factor explaining compositional variation, followed by inclination and proportion of exposed rock surface. Despite its endangered status, E. aurantiacum showed a relatively high degree of ecological plasticity across these semiarid grassland habitats that probably allows it to survive in different environments, including in some cases those impacted by human activities. This prompts for habitat‐ more than species‐level conservation actions. According to their characteristics and threats, habitat‐specific management practices are recommended for long term conservation of plant species communities in the different ecological niches.     

Role of cytochrome P450 2B sequence variation and gene copy number in facilitating dietary specialization in mammalian herbivores

Theory postulates that dietary specialization in mammalian herbivores is enabled by a specialized set of liver enzymes that process the high concentrations of similar plant secondary metabolites (PSMs) in the diets of specialists. To investigate whether qualitative and quantitative differences in detoxification mechanisms distinguish dietary specialists from generalists, we compared the sequence diversity and gene copy number of detoxification enzymes in two woodrat species: a generalist, the white‐throated woodrat (Neotoma albigula) and a juniper specialist, Stephens’ woodrat (N. stephensi). We focused on enzymes in the cytochrome P450 subfamily 2B (CYP2B), because previous research suggests this subfamily plays a key role in the processing of PSMs. For both woodrat species, we obtained and sequenced CYP2B cDNA, generated CYP2B phylogenies, estimated CYP2B gene copy number and created a homology model of the active site. We found that the specialist possessed on average ~5 more CYP2B gene copies than the generalist, but the specialist's CYP2B sequences were less diverse. Phylogenetic analysis of putative CYP2B homologs resolved woodrat species as reciprocally monophyletic and suggested evolutionary convergence of distinct homologs on similar key amino acid residues in both species. Homology modelling of the CYP2B enzyme suggests that interspecific differences in substrate preference and function likely result from amino acid differences in the enzyme active site. The characteristics of CYP2B in the specialist, that is greater gene copy number coupled with less sequence variation, are consistent with specialization to a narrow range of dietary toxins.     

Determinants of parasitoid communities of willow‐galling sawflies: habitat overrides physiology,host plant and space

Studies on the determinants of plant–herbivore and herbivore–parasitoid associations provide important insights into the origin and maintenance of global and local species richness. If parasitoids are specialists on herbivore niches rather than on herbivore taxa, then alternating escape of herbivores into novel niches and delayed resource tracking by parasitoids could fuel diversification at both trophic levels. We used DNA barcoding to identify parasitoids that attack larvae of seven Pontania sawfly species that induce leaf galls on eight willow species growing in subarctic and arctic–alpine habitats in three geographic locations in northern Fennoscandia, and then applied distance‐ and model‐based multivariate analyses and phylogenetic regression methods to evaluate the hierarchical importance of location, phylogeny and different galler niche dimensions on parasitoid host use. We found statistically significant variation in parasitoid communities across geographic locations and willow host species, but the differences were mainly quantitative due to extensive sharing of enemies among gallers within habitat types. By contrast, the divide between habitats defined two qualitatively different network compartments, because many common parasitoids exhibited strong habitat preference. Galler and parasitoid phylogenies did not explain associations, because distantly related arctic–alpine gallers were attacked by a species‐poor enemy community dominated by two parasitoid species that most likely have independently tracked the gallers’ evolutionary shifts into the novel habitat. Our results indicate that barcode‐ and phylogeny‐based analyses of food webs that span forested vs. tundra or grassland environments could improve our understanding of vertical diversification effects in complex plant–herbivore–parasitoid networks.     

Expression of biotransformation genes in woodrat (Neotoma) herbivores on novel and ancestral diets: identification of candidate genes responsible for dietary shifts

The ability of herbivores to switch diets is thought to be governed by biotransformation enzymes. To identify potential biotransformation enzymes, we conducted a large-scale study on the expression of biotransformation enzymes in herbivorous woodrats ( Neotoma lepida ). We compared gene expression in a woodrat population from the Great Basin that feeds on the ancestral diet of juniper to one from the Mojave Desert that putatively switched from feeding on juniper to feeding on creosote. Juniper and creosote have notable differences in secondary chemistry, and thus, should require different biotransformation enzymes for detoxification. Individuals from each population were fed juniper and creosote diets separately. After the feeding trials, hepatic mRNA was extracted and hybridized to laboratory rat microarrays. Hybridization of woodrat samples to biotransformation probes on the array was 87%, resulting in a total of 224 biotransformation genes that met quality control standards. Overall, we found large differences in expression of biotransformation genes when woodrats were fed juniper vs. creosote. Mojave woodrats had greater expression of 10× as many biotransformation genes as did Great Basin woodrats on a creosote diet. We identified 24 candidate genes that may be critical in the biotransformation of creosote toxins. Superoxide dismutase, a free radical scavenger, was also expressed to a greater extent by the Mojave woodrats and may be important in controlling oxidative damage during biotransformation. The results are consistent with the hypothesis that biotransformation enzymes limit diet switching and that woodrats in the Mojave have evolved a unique strategy for the biotransformation of creosote toxins.     

Diversity and novelty of the gut microbial community of an herbivorous rodent (Neotoma bryanti)

Mammalian herbivores host diverse microbial communities to aid in fermentation and potentially detoxification of dietary compounds. However, the microbial ecology of herbivorous rodents, especially within the largest superfamily of mammals (Muroidea) has received little attention. We conducted a preliminary inventory of the intestinal microbial community of Bryant’s woodrat (Neotoma bryanti), an herbivorous Muroidea rodent. We collected woodrat feces, generated 16S rDNA clone libraries, and obtained sequences from 171 clones. Our results demonstrate that the woodrat gut hosts a large number of novel microorganisms, with 96% of the total microbial sequences representing novel species. These include several microbial genera that have previously been implicated in the metabolism of plant toxins. Interestingly, a comparison of the community structure of the woodrat gut with that of other mammals revealed that woodrats have a microbial community more similar to foregut rather than hindgut fermenters. Moreover, their microbial community was different to that of previously studied herbivorous rodents. Therefore, the woodrat gut may represent a useful resource for the identification of novel microbial genes involved in cellulolytic or detoxification processes.     

Pup development and maternal behavior in captive Key Largo woodrats (Neotoma floridana smalli)

This study describes the first systematic observations of maternal behavior and pup development of captive Key Largo woodrats (Neotoma floridana smalli) during the first 30 days of life. Data were collected on six litters of pups born to four dams between December 2006 and July 2007. Gestations for the six litters averaged 38 days and all dams exhibited adequate maternal care postpartum. Key Largo woodrat maternal and pup behavior was generally consistent with behavior observed in other woodrat species. We observed greater pup independence from the dam and a marked change in social interactions between days 13–22. No sex differences in pup development or maternal care were observed. Activity budgets were consistent across dams and across days within the observation period. Although dams spent much of their time inactive with pups attached to their teats, the average percent of intervals with at least one pup observed attached decreased steadily during the 30‐day observation period. Attachment of pups to the dams' teats did not interfere with dams' ability to forage. Feeding with pups attached and feeding following active detachment of pups were both common. Dams were observed to actively detach pups by performing a circular turning motion. This information has application for the future management of this endangered species in captivity and in the wild. Zoo Biol 27:394–405, 2008. © 2008 Wiley‐Liss, Inc.     

Herbivore control in connected seascapes: habitat determines when population regulation occurs in the life history of a key herbivore

Herbivore outbreaks often trigger catastrophic overgrazing events in marine macrophyte ecosystems. The sea urchin Paracentrotus lividus, the dominant herbivore of shallow Mediterranean seascapes, is capable of precipitating shifts to barrens when its populations explode. Paracentrotus lividus is found ubiquitously in rocky macroalgal communities and in sandy seagrass meadows of Posidonia oceanica, two of the most important subtidal habitats in the Mediterranean. We explored if habitat‐specific regulation across the principal stages of the urchin life cycle could help explain the persistence of these populations in connected mosaics. We measured each of three relevant ecological processes (i.e. settlement, post‐settlement survival and predation) across a wide stretch of the Mediterranean coast (ca 600 km). Our results show that habitat‐specific regulation is critical in determining urchin populations: each habitat limited urchin sub‐populations at different life stages. Settlement was never limiting; urchins settled at similar rates in both habitats across the coast. Post‐settlement survival was a clear bottleneck, particularly in seagrass meadows where no juvenile urchins were recorded. Despite this bottleneck in seagrasses, adult urchin populations were very similar in both seagrass and macroalgal habitats indicating that other processes (potentially migration) could be key in determining adult distributions across the mosaic. The fact that population regulation is clearly habitat‐specific suggests that sea urchin populations may be significantly buffered from bottlenecks in mixed seascapes where both habitats co‐occur. Sea urchin populations can therefore persist across the seascape despite strong habitat‐specific regulation either by maintaining reproductive output in one habitat or by migrating between them. By affording these regulatory escapes to habitat‐modifying species, patchy mosaics may be much more prone to herbivore outbreaks and a host of cascading effects that come in their wake.     

Intraspecific and interspecific competition induces density‐dependent habitat niche shifts in an endangered steppe bird

Interspecific competition is a dominant force in animal communities that induces niche shifts in ecological and evolutionary time. If competition occurs, niche expansion can be expected when the competitor disappears because resources previously inaccessible due to competitive constraints can then be exploited (i.e., ecological release). Here, we aimed to determine the potential effects of interspecific competition between the little bustard (Tetrax tetrax) and the great bustard (Otis tarda) using a multidimensional niche approach with habitat distribution data. We explored whether the degree of niche overlap between the species was a density‐dependent function of interspecific competition. We then looked for evidences of ecological release by comparing measures of niche breadth and position of the little bustard between allopatric and sympatric situations. Furthermore, we evaluated whether niche shifts could depend not only on the presence of great bustard but also on the density of little and great bustards. The habitat niches of these bustard species partially overlapped when co‐occurring, but we found no relationship between degree of overlap and great bustard density. In the presence of the competitor, little bustard's niche was displaced toward increased use of the species' primary habitat. Little bustard's niche breadth decreased proportionally with great bustard density in sympatric sites, in consistence with theory. Overall, our results suggest that density‐dependent variation in little bustard's niche is the outcome of interspecific competition with the great bustard. The use of computational tools like kernel density estimators to obtain multidimensional niches should bring novel insights on how species' ecological niches behave under the effects of interspecific competition in ecological communities.