The diets of spiders in ginger sites contain more non-native taxa
The metabarcoding approach used in our study provided an avenue for assignment of arthropod taxa to native versus invasive status for any taxa included in the GenBank database. For those not included, identifications were only to genus, family, or order. However, assignment to native versus nonnative status is simplified for the Hawaiian arthropods because many taxa are endemic at the genus level (Eldredge & Miller 1995) and many insect families and orders do not occur in the native arthropod biota, allowing classification at a higher taxonomic level (Howarth 1990).
Our results showed that native taxa were found in the diet of the endemic Pagiopalus in both native forest and ginger-invaded sites: Campsicnemus (Dolichopodidae) in 10 spiders from ginger sites and 3 from native forest, Limonia (Limoniidae) in 10 spiders from ginger sites and 4 from native forest, andNesophrosyne (Cicadellidae) from 7 spiders from ginger sites and 13 spiders from native forest. These taxa are included within some of the largest insect radiations in the islands. The hemipterans in family Cicadellidae are specific to native plant species (Bennett, pers comm, Bennett & O’Grady 2012). As the dominant prey family in the diets of spiders collected in native forest (detected in 25 of the 58 spiders), and as endemic taxa associated with native flora, these hemipterans likely represent an important prey source with which Pagiopalusevolved. Because the host plants are absent from heavily invaded ginger sites, access to host-specific native prey is likely limited for spiders in ginger forest.
Differences between the diets of Pagiopalus in ginger and native-forest sites were driven largely by the high numbers of nonnative taxa (Figure 7), specifically Entomobryomorpha (Collembola) which was detected in the diets of 28 spiders from ginger while being almost entirely absent in the diets of spiders from the native sites (Figure 5). Collembolans have been found to greatly outnumber other arthropods in ginger-invaded habitat (unpublished data). The collembolanSalina celebensis was the most prevalent in ginger (detected in 18 spiders) and was not detected in native forest. This species, introduced from Asia, is characteristic of moist understory vegetation, with its extraordinary abundance noted in previous studies (Christiansen & Bellinger 1994; Gruner & Taylor 2006; Gruner et al. 2005). The other species of Collembola was Tomocerus sp ; although not identifiable to species, T. minor , introduced from Europe has been known from Hawaii, for > 50 years (Christiansen & Bellinger 1992). Tomocerus was found in a single spider in native forest contrasting its detection in 15 spiders from ginger sites.
The much higher numbers of Collembola in invaded habitat may enhance the survival of native spiders such as Pagiopalus , simply because of their abundance; alternatively, they may serve to detract from their survival if they do not support the nutritional needs of the predator. Previous work has shown that, while being a common source of prey in cursorial spiders (Birkhofer & Wolters 2012), Collembola have mixed nutritional benefits. Studies have found inclusion of certain species of collembolans (Tomocerus bidentatus, Isotoma anglicana) in the diets of spiders increases reproductive output or survival while other species (Folsomia candida , Folsomia fimetaria, Isotoma trispinata) drastically decrease reproductive output or result in high mortality (Møller Marcussen et al. 1999; Toft & Wise 1999; Rickers et al. 2006). These studies additionally note the possible toxicity of certain collembolans, resulting in a significant fitness cost (Toft & Wise 1999). Generalist predators are well documented to be dietarily selective when given a diverse set of prey, preferentially eating prey items which have the highest nutritional benefit (Michalko et al. 2019; Rendon et al. 2019; Toft 1999). In order to obtain nutritional balance, however, a mixed diet consisting of nutritionally imbalanced or even toxic prey may be consumed when optimal prey is in low abundance. Supplementation of Collembola and other prey not found in native forest may be a necessity for spiders in ginger because of the lack of access to host-plant specific taxa. The higher prey diversity and higher reads in spiders from ginger sites supports this, showing a mixed dietary strategy not adopted by spiders in native forest. While prey may be available in ginger, the prey itself may be less suitable than native insects found predominantly in native forest; high prevalence of some prey items, in particular Collembola, may create an ecological trap by attracting spiders while requiring spiders to predate more taxa to accommodate nutritional inefficiencies.