Assessing the impact of the threatened crucian carp (Carassius carassius) on pond invertebrate diversity: A comparison of conventional and molecular tools

Fishes stocked for recreation and angling can damage freshwater habitats and negatively impact biodiversity. The pond‐associated crucian carp (Carassius carassius) is rare across Europe and is stocked for conservation management in England, but its impacts on pond biota are understudied. Freshwater invertebrates contribute substantially to aquatic biodiversity, encompassing many rare and endemic species, but their small size and high abundance complicate their assessment. Practitioners have employed sweep‐netting and kick‐sampling with microscopy (morphotaxonomy), but specimen size/quality and experience can bias identification. DNA and environmental DNA (eDNA) metabarcoding offer alternative means of invertebrate assessment. We compared invertebrate diversity in ponds (N = 18) with and without crucian carp using morphotaxonomic identification, DNA metabarcoding and eDNA metabarcoding. Five 2 L water samples and 3 min sweep‐net samples were collected at each pond. Inventories produced by morphotaxonomic identification of netted samples, DNA metabarcoding of bulk tissue samples and eDNA metabarcoding of water samples were compared. Alpha diversity was greatest with DNA or eDNA metabarcoding, depending on whether standard or unbiased methods were considered. DNA metabarcoding reflected morphotaxonomic identification, whereas eDNA metabarcoding produced markedly different communities. These complementary tools should be combined for comprehensive invertebrate assessment. Crucian carp presence minimally reduced alpha diversity in ponds, but positively influenced beta diversity through taxon turnover (i.e., ponds with crucian carp contained different invertebrates to fishless ponds). Crucian carp presence contributes to landscape‐scale invertebrate diversity, supporting continued conservation management in England. Our results show that molecular tools can enhance freshwater invertebrate assessment and facilitate development of more accurate and ecologically effective pond management strategies.

identification (Briers & Biggs, 2003  In this study, we assessed invertebrate diversity in UK ponds with and without crucian 131 carp using metabarcoding in conjunction with morphotaxonomic identification. The entire 132 taxon inventories generated by each monitoring tool (standard methods) were compared to 133 evaluate which tool provided the most holistic assessment of invertebrate diversity. We also 134 compared the taxon inventories produced by each method when potential biases were 135 removed, i.e. taxa with no reference sequences and meiofauna that would not be captured by a 136 1 mm mesh net (unbiased methods). The effect of crucian carp presence on invertebrate 137 diversity was then determined using the taxon inventories from the standard or unbiased 138 methods individually and combined. We hypothesised that: 1) DNA metabarcoding and 139 morphotaxonomic identification would produce congruent invertebrate communities (e.g. 2) DNA and eDNA metabarcoding would enable species resolution data for problematic taxa 143 that cannot be morphotaxonomically identified to species-level using standard keys (e.g.  fishes that occurred in three and one of the Norfolk ponds respectively, but at much lower 168 density (see Table S1). All study ponds were selected to be similar in terms of morphometry and      constructed for DNA and eDNA metabarcoding using a two-step PCR protocol (Appendix 1).

259
During the first PCR (performed in triplicate), the target region was amplified using 260 oligonucleotides comprising the locus primers, sequencing primers, and pre-adapters. DNA 261 from the exotic, terrestrial two-spotted assassin bug (Platymeris biguttatus) was used for PCR 262 positive controls (tissue DNA n = 9, eDNA n = 11) as this species is not found in the UK, whilst 263 sterile molecular grade water (Fisher Scientific UK Ltd, UK) substituted template DNA for PCR 264 negative controls (tissue DNA n = 9, eDNA n = 11). PCR products were individually purified using  The DNA and eDNA libraries were sequenced on an Illumina® MiSeq using 2 x 300 bp V3 281 chemistry with 10% and 20% PhiX Sequencing Control respectively (Illumina, Inc, CA, USA).

282
Raw sequences were converted to taxonomic assignments using metaBEAT v0.97.11, a 283 custom pipeline for reproducible analysis of metabarcoding data (https://github.com/HullUni-284 bioinformatics/metaBEAT). After quality filtering, trimming, merging, chimera detection, and 285 clustering, non-redundant query sequences were compared against our reference databases 286 using BLAST (Zhang, Schwartz, Wagner, & Miller, 2000). Putative taxonomic identity was 287 assigned using a lowest common ancestor (LCA) approach based on the top 10% BLAST matches 288 for any query matching with at least 90% identity to a reference sequence across more than 289 11 80% of its length. Unassigned sequences were subjected to a separate BLAST against the 290 complete NCBI nucleotide (nt) database at 90% identity to determine the source via LCA as   family inventories produced by each method (hereafter standard methods) were compared. 307 We also compared inventories from each method after correcting for known biases: taxa 308 without reference sequences (i.e. those which cannot be detected with metabarcoding) were 309 removed from the morphotaxonomic dataset, and meiofauna (i.e. taxa that would escape a 1 310 mm mesh net) were removed from the metabarcoding datasets (hereafter unbiased methods).

311
The effect of crucian carp on invertebrate diversity was then assessed at species and family-  species-level analyses, only species records were used, whereas family-level analyses included 317 species, genus and family records grouped at the taxonomic rank of family. 318 We define alpha diversity as taxon richness (species or families) within individual ponds, 319 and beta diversity as the difference between communities present at each pond whilst 320 accounting for taxon identity (Baselga & Orme, 2012). Jaccard dissimilarity was used as a 321 measure of beta diversity for all analyses. For each standard or unbiased dataset, the following 322 analyses were performed. Alpha diversity (response variable) was obtained using the 323 specnumber function in vegan v2.5-6 and modelled against sampling method (explanatory 324 variable), then modelled against crucian carp presence-absence (explanatory variable). Using Pairwise Tukey's HSD tests were used to assess significance of differences. We used betapart    (Table 1).

15
Conversely, unbiased sampling method influenced alpha diversity of ponds at species-level 404 (GLM χ 2 2 = 14.614, P < 0.001), but not family-level (GLM χ 2 2 = 5.214, P = 0.074). Significant 405 differences between the species-level alpha diversity means of morphotaxonomic identification 406 and DNA metabarcoding, and DNA and eDNA metabarcoding were observed (Table 1). With 407 standard methods, species and family-level alpha diversity was highest using eDNA  (Figs. 2bi, bii). For beta diversity, MVDISP did not significantly differ between 412 standard or unbiased methods for turnover, nestedness-resultant or total beta diversity at 413 either taxonomic rank (Table S8). Both standard and unbiased sampling method had moderate 414 and strong positive effects on turnover and total beta diversity at species (Figs. 3ai, aiii, ci, ciii) 415 and family-level (Figs. 3aii, aiv, cii, civ) respectively, but not nestedness-resultant (Table S9). Here, we report the impact of crucian carp stocking assessed using standard methods as these 420 provide the most comprehensive evaluation of invertebrate diversity. Crucian carp impact 421 assessment using unbiased methods is reported in the Supporting Information (Appendix 2;

435
Total beta diversity of ponds was consistently high at species and family-level for 436 independent and combined methods (standard or unbiased). Variation in invertebrate 437 community composition was driven by turnover rather than nestedness-resultant (Tables 3, 438 S11). MVDISP differed between ponds for species-level total beta diversity with eDNA 439 metabarcoding, and species and family-level total beta diversity with all methods combined.

440
Fishless ponds had significantly lower dispersion than ponds with crucian carp (Table S14).

452
Additional analyses undertaken on the combined method datasets supported an effect 453 of crucian carp presence and excluded the influence of abiotic variables on invertebrate 454 diversity. At species-level, only pond area was identified by forward selection as a significant 455 abiotic variable for turnover and total beta diversity. Variance partitioning analysis was 456 undertaken for turnover and total beta diversity using crucian carp presence-absence and pond 457 area. Using the adjusted R 2 values, biotic and abiotic variables explained 7.04% and 4.75% of 458 the total variation in species turnover and total beta diversity respectively (Fig. 6). Crucian carp 459 presence-absence significantly contributed to turnover ( Fig. 6a; adjusted R 2 = 5.55%, F1 = 2.031, 460 P = 0.001) and total beta diversity ( Fig. 6b; adjusted R 2 = 4.05%, F1 = 1.730, P = 0.001), whereas 461 17 pond area explained less variance and did not influence turnover ( Fig. 6a; adjusted R 2 = 2.94%, 462 F1 = 0.340, P = 0.852) or total beta diversity ( Fig. 6b; adjusted R 2 = 1.93%, F1 = 1.118, P = 0.218).   Critically, we used a 1 mm mesh net which is designed to sample macroinvertebrates, 489 whereas a 500 μm mesh net would also sample the meiofauna detected by metabarcoding.   proportions of invertebrate taxa, thus different fishes will suppress numbers of and confer 642 benefits to different invertebrate taxa (Batzer, Pusateri, & Vetter, 2000). Invasive fish species 643 may be more detrimental than non-invasive species, for example, the mosquitofish (Gambusia 644 affinis) reduced zooplankton abundance and macroinvertebrate density by 90% and 50% 645 respectively after introduction in a wetland ecosystem experiment (Preston et al., 2017). 646 Additionally, invertebrate species richness and abundance was found to decrease as duration of 647 stocking increased (Schilling et al., 2009a). Therefore, local-and regional-scale diversity may 648 benefit most from pond mosaics that maintain fish-free and fish-containing (at low-moderate