The combined effects of multiple invasive species on persistence of imperiled Pahrump Poolfish

Many ecosystems have been invaded by more than one non-native species, but research evaluating the combined effects of multiple invasive species has been limited. In the southwest USA, many aquatic systems have been invaded by multiple species such as non-native crayfish and non-native fishes. We used experimental mesocosms to test individual and combined effects of invasive Red Swamp Crayfish, Procambarus clarkii, and Western Mosquitofish, Gambusia affinis, on endangered Pahrump Poolfish, Empetrichthys latos. We found that crayfish alone reduced adult poolfish survival; however, crayfish by themselves had moderate but non-significant impacts on the production of poolfish juveniles. Mosquitofish had no effect on the survival of poolfish adults, but significantly reduced recruitment of juveniles. When both crayfish and mosquitofish were present, both adult survival and juvenile production were significantly decreased. These findings were consistent with the recent decline of a wild poolfish population from over 10,000 fish to less than 1,000 poolfish following the establishment of crayfish and mosquitofish. This study demonstrates that conservation management of the Pahrump Poolfish must have active management and removal of invasive species, otherwise extirpation and eventually extinction will likely occur. This study also provides an example of the compounding effects of multiple invasive species.

Methods
Western Mosquitofish were obtained from Sutter-Yuba Mosquito and Vector Control district in Yuba City, CA. Poolfish used in this experiment included a mixture of wild poolfish and lab-reared poolfish. The wild poolfish were collected from on 13 June 2017 from Shoshone Stock Pond (White Pine County, NV) while the lab-reared poolfish were descendants from poolfish originally collected in 2014 from Spring Mountain Ranch State Park, Clark County [31]. Red Swamp Crayfish were sourced from Carolina Biological suppliers Burlington, NC.

 Three fish communities were used in this experiment forming a single block within a randomized block design. Each of seven blocks contained a total of three mesocosms including the following treatments: I.) allopatric poolfish, II.) poolfish sympatric with crayfish, and III.) poolfish sympatric with both mosquitofish and crayfish. We did not include a poolfish + mosquitofish treatment because of the limited number of poolfish available for this experiment.  Further, three previous experiments consistently showed that mosquitofish effectively eliminated the production of juvenile poolfish.

Each block of three tanks was replicated seven times for a total of 21 experimental tanks, arranged in a linear sequence. All 21 tanks received seven adult poolfish of indeterminate sex and of indeterminate population of origin (Shoshone Stock Pond 2017 or Spring Mountain Ranch 2014). Four individual crayfish were introduced into two randomly selected mesocosms per block. One of the two crayfish mesocosms within each block was randomly selected to receive mosquitofish, including five gravid females and two males. Crayfish density was maintained by replacing any crayfish that died.

All mesocosms were provided with reclaimed PVC vinyl Fishiding® structures to simulate aquatic plants and to provide spatial structure along with ~57 L of river rock. Supplemental food was provided every day to each tank at rates of ~2-3% of total fish biomass, while crayfish were fed twice weekly. Food consisted of Tetra tropical flake, and Aquatic Arts (Fish, Inverts, and Aquatic Plant) sinking pellets. Water quality was assessed weekly for ammonia and nitrates. All tanks were checked daily for mortalities, and to ensure air flow was constant from air stones. After ten-weeks, the tanks were drained, and juveniles and adults were counted.

Data were analyzed using JMP Pro 17® software. We tested for treatment effects on adult survival, juvenile production, and the number of juveniles produced per surviving adult poolfish. We used ANOVA followed by Tukey HSD while maintaining experimental-wise alpha at 0.05. For each test, block was not significant and was eliminated from the final model. Non-parametric analyses produced the same significance levels among treatments for all tests, but here we report the parametric ANOVA results.