Native fish translocations mediated by anthropogenic drainage modifications in southern New Zealand

ABSTRACT

Freshwater-limited fish populations are often tightly constrained by river drainage boundaries. As a case in point, the distribution of lineages within New Zealand’s diverse Galaxias vulgaris complex is broadly structured by geographic barriers, reflecting tectonic processes. However, several drainages of the Central Otago region have been locally modified by past gold-mining activities, with artificial water races connecting formerly isolated headwaters of distinct river drainage systems. Here we synthesise published genetic data to highlight the role of anthropogenic catchment modification in redistributing fish diversity. These data show that several local phylogeographic anomalies for stream-resident Galaxias fishes are closely linked to anthropogenic connections across major drainage divides. While these anthropogenic translocations may parallel natural geologically-driven mixing events that occur on deeper time scales, they nevertheless have potential conservation implications given the increasing fragmentation of native fish populations. Our review also highlights the role of sub-catchment units in shaping the hierarchical structure of intraspecific biodiversity.

KEYWORDS:

• Biogeography
• conservation
• freshwater
• genetics
• river

Introduction

Freshwater-limited fish assemblages often exhibit high levels of species diversity across relatively small geographic scales (Adams et al. 2014; Raadik 2014; Shelley et al. 2018; Ronco et al. 2021). This diversity partly reflects the biogeographic constraints imposed by drainage boundaries (Waters et al. 2020), with populations unable to disperse beyond their natal catchments. Additionally, dynamic geological processes (Waters et al. 2001, 2020; Craw et al. 2016b) can rapidly reorient drainage boundaries, and thereby contribute further to the diversification of freshwater-limited lineages (Barluenga et al. 2006; Melo et al. 2021; Ronco et al. 2021).

The family Galaxiidae represents an important component of the Southern Hemisphere’s freshwater fish fauna (McDowall 1970, 1990; Burridge et al. 2012). In recent years, genetic analyses have revealed several independent species-rich radiations of stream-resident galaxiids in southern landmasses (Australia: Adams et al. 2014; Raadik 2014; New Zealand: Allibone et al. 1996; Waters and Wallis 2001a, 2001b; Africa: Chakona et al. 2013; South America: Delgado et al. 2019). In New Zealand’s South Island, for instance, the ‘common river galaxias’ (e.g. McDowall and Wallis 1996; McDowall 1997), once considered a single species (Galaxias vulgaris sensu lato; McDowall 1970, 1990), is now recognised as comprising 12 taxa, only six of which have been formally described (Waters et al. 2010; Burridge et al. 2012; Campbell et al 2022). In the context of widespread habitat degradation, and population fragmentation following the introduction of salmonid predators, conserving such newly-recognised freshwater biodiversity can present major conservation challenges (McDowall 2006; Dunn et al. 2018).

The Otago region of southern South Island (Figure 1) represents the ‘centre’ of Galaxias vulgaris complex diversity (Figure 1A,B), with the Clutha and Otago river systems both housing disproportionately large numbers of stream-resident taxa (Waters et al. 2001, 2010, 2015). This southern region is thought to represent the geographic ‘origin’ of the radiation (‘out of the south’: McDowall and Wallis 1996; Waters and Wallis 2001b), with diversification apparently underpinned by a dynamic geological history over the past 5 million years. Specifically, mountain building and drainage reorientation events have driven isolation, diversification and redistribution of fish lineages among river systems (Figure 1A,B; 2A–C; Burridge et al. 2006, 2007, 2008; Craw et al. 2016a; Waters et al. 2020).

Figure 1. New Zealand’s South Island is home to a diverse assemblage of freshwater-limited fish taxa, the Galaxias vulgaris complex, including a variety of A, ‘roundhead’ and B, ‘flathead’ lineages whose distributions are broadly constrained by major river drainage boundaries. In many parts of the Otago region C, drainage patterns were anthropogenically modified by gold mining operations, with water-races linking numerous headwaters. While most of these artificial connections redistributed water only within river systems (yellow circles), in several cases such connections linked the headwaters of distinct river systems (red), leading to potential translocation and mixing between anciently-diverged species/populations.

In addition to these long-term geological processes, the Otago area also has a dynamic history of anthropogenic drainage modification associated with the Otago gold rush starting in the 1860s. Across the region (Figure 1C), gold-mining operations in the 1800s often relied on construction of artificial channels, locally called ‘water races’, to supply water. Some of these water races passed across drainage divides (Figure 1C; 2A), potentially connecting previously isolated biological lineages.

In this paper, we synthesise published evidence for trans-divide dispersal of stream-resident galaxiid fish mediated by water races (Figure 2A), focussing on phylogeographic data (Figure 2B). A previous genetic study has suggested that artificial headwater connections via water-race construction may have led to redistribution of species among river catchments in the Otago region (Esa et al. 2000). However, genome-wide analyses, preferably with geographic replication, are required to more rigorously assess the conservation genetic effects of such anthropogenic drainage modifications. Here we examine more broadly the role of water races as agents of anthropogenic freshwater biological translocation, using assessment of geographical and fish phylogeographic data as outlined schematically in Figure 2.

Figure 2. Schematic depiction of hypothetical biogeographic A, and phylogeographic B, signatures of recent anthropogenic translocation across a major drainage divide mediated by water race (WR) connections. The translocated lineage (indicated with stars) represents a biogeographic anomaly within its new river system A, and retains strong phylogenetic similarity B with lineages from its original source tributary. Given the recent anthropogenic timeframe, the translocated and source populations are unlikely to have evolved reciprocal monophyly B.

Methods

The geographical component of this review is based on a combination of published records, topographic map data, satellite imagery, and our direct field observations. Early gold mining in Otago focussed primarily on placer deposits, where detrital gold was concentrated in surficial gravels. The gravels were excavated and gold extracted using gravity-driven water flow. Hence, miners required large volumes of water with sufficient topographic fall to provide the energy for the mining operations. Most of these mining operations obtained this water by constructing water races from the upper reaches of nearby streams within the same catchment (Figure 1, orange dots). However, where nearby streams were too small, water was obtained from adjacent river catchments, with water races crossing the topographic divides between them. Seven water races connected headwaters of completely distinct catchments (Figure 1). These races were major construction and maintenance operations at the time, in steep mountainous landscapes with periodic major rain events and landslides. We focus in particular on three water races (Figure 3) that connected separate catchments and lie in regions that have been the focus of detailed fish-genetic analyses (Campbell et al 2022).

Figure 3. Examples of anthropogenic water-race connections (red) linking headwaters across major drainage divides (dashed lines) in southern New Zealand, constructed for past gold mining. Water race links are documented between headwaters of A, Waitahuna (Clutha) and Tokomairaro rivers (Barnett 2016); B, Waipori (Taieri) and Tuapeka (Clutha) rivers; and C, Totara Ck (Taieri) and Pool Burn (Clutha). D, Annotated Google Earth image of the race in C, as it crosses the divide. Note rotated view compared to map in C.

The longest of these three water races connected headwaters of the Tokomairaro River with those of the adjacent Waitahuna River (Clutha catchment; Figure 3A and 4A). Water flowed southwest to service a long-lived gold mining operation in a tributary of the Waitahuna River (Figure 3A). The continuity of the race declined in the middle of the twentieth Century after the mining ceased. The race from the headwaters of the Waipori River (Taieri catchment) to the headwaters of a Tuapeka River tributary (Clutha catchment) was developed to service mining operations in the Tuapeka River area (Figure 3B) that was the site of the original Otago gold rush in 1861. Initial mining in that area used waters collected from within the Tuapeka catchment, but when these became insufficient as the scale of mining progressed, additional water was collected from the Waipori River (Figure 3B).

The third divide-crossing water race example in our study, on Rough Ridge (Figure 1C; Esa et al. 2000) was initially developed to service a mining area on a remote upland plateau (Figure 3C). The area has complex topography and drainage patterns, and water was gathered from local catchments as well as the larger race that crossed the drainage divide and carried water southwards (Figure 3C,D). Subsequently, the drainage and water race systems were further developed in association with a large water storage lake for agricultural irrigation, and the water race now flows across the divide towards the north (Figure 3C,D). These headwater races even today house substantial populations of Galaxias fishes (Esa et al. 2000).

Phylogeographic criteria

To assess biological evidence for possible anthropogenic translocation events, we review phylogeographic data for South Island stream-resident Galaxias fishes (Campbell et al. 2022). We rely on recently-published phylogenetic analyses of single-nucleotide polymorphism (SNP) data obtained via single-digest Genotyping-by-Sequencing (GBS; Baird et al. 2008; Elshire et al. 2011) from a study assessing phylogeographic relationships of Otago’s stream-resident fish populations (Campbell et al. 2022).

Intraspecific diversity of the stream-resident Galaxias vulgaris complex is mostly structured according to major river drainage boundaries (Waters et al. 2001, 2015), with different rivers and tributaries often hosting distinct lineages (Campbell et al. 2022). In this context, the occasional detection of distinctive lineages distributed across major drainage divides could reflect either ‘ancient’ geologically-mediated processes (typically >10,000 years ago; and up to several hundred thousand years ago; Waters et al. 2001; Burridge et al. 2006, 2007; Waters et al. 2020b), or relatively recent anthropogenic (<750 years ago; e.g. Esa et al. 2000; Waters et al. 2002) range shifts. The key predictions of anthropogenic range shifts include (i) the anomalous distribution of distinctive stream-resident clades across major drainage divides; and (ii) unusually shallow genetic divergences among neighbouring headwater populations (Figure 2). The translocation scenarios outlined in our review highlight cases where phylogenomic and biogeographic anomalies (Campbell et al. 2022) satisfy the two criteria outlined above, and are spatially linked to trans-divide water races.

Genetic evidence of fish translocation

Waitahuna-Tokomairaro translocation

The ‘roundhead’ lineage G. ‘Pomahaka’ is widespread across low-gradient streams of the lower Clutha system (Figure 4A), with an additional outlying record from the adjacent Tokomairaro catchment (XT; Figure 4A). Genome-wide analyses of this taxon have revealed strong within-drainage phylogeographic structure, with distinct Clutha sub-catchments (e.g. Waitahuna, Pomahaka, Waipahi) housing monophyletic sub-clades (Figure 4A). Overall, it is clear that main-channel genetic connectivity within this taxon is minimal, with subcatchments comprising discrete populations (see also Waters and Burridge 2016). This apparent lack of gene flow is further emphasised by the finding that multiple samples from single localities almost inevitably form monophyletic subclades (phylogeny in Figure 4A), even among proximal locations within subcatchments (e.g. G. ‘Pomahaka’ samples XO, XH, XA, XB, XK).

Figure 4. Phylogenomic evidence for translocation of native fish lineages (red arrows) across major river drainage divides (dashed lines) in southern New Zealand, associated with anthropogenic water-race connections (Figure 3). Likely translocation events (indicated with red stars) are inferred from a genome-wide phylogenetic analysis of Galaxias (Campbell et al. 2022), using phylogeographic criteria outlined in Figure 2. Recent, trans-divide dispersal between headwater populations is inferred for A, Galaxias ‘Pomahaka’ in the Waitahuna (Clutha) and Tokomairaro (Manuka) drainages; B, Galaxias pullus in Waipori (Taieri) and Tuapeka (Clutha) river systems. Further north C, in the Rough Ridge region, the distribution of G. ‘sp D’ (Clutha) genotypes within the range of G. depressiceps (Taieri) represents a major biogeographic anomaly (see Figure 2) in close proximity to a water race linking headwaters of Totara Creek and Pool Burn (Figure 3C,D). The trans-divide distributions of mtDNA clades (Esa et al. 2000; Burridge et al. 2007; Waters et al. 2015) and phylogenomic (Campbell et al. 2022; circles with black outlines) lineages in this region reveal a complex history of anthropogenic translocation and introgression (asterisks). On the eastern side of the divide, gold diggings have extended the introgression of G. ‘sp D’ alleles into the headwaters of a second Taieri tributary: Waimonga Creek (Esa et al. 2000).

In the context of this strong phylogeographic structure within and among streams, the close phylogenetic relationship between the ‘outlier’ Tokomairaro R (Manuka; XT) record of G. ‘Pomahaka’ and nearby populations from headwaters of the adjacent Waitahuna (Clutha system; XG, XW, XD) subcatchment (Figure 4A) suggests a potential headwater exchange event (Figure 2A). In particular, the sister relationship specifically between Manuka (XT) and upper Waitahuna samples (XG, XW), even to the exclusion of lower Waitahuna (XD) samples, supports recent gene flow across the low divide separating these drainages (Figure 4A). The presence of an anthropogenic water race connection across the divide in precisely this region provides a clear mechanistic explanation for such a recent translocation (Figure 3A), and thus likely explains the anomalous detection of this lineage within the Tokomairaro system. The long branch length of the Manuka sample XT (Figure 4A) likely reflects rapid genetic drift linked to a recent founder event. Future fine-scale sampling of both Waitahuna and Tokomairaro systems promises to shed additional light on the translocation dynamics and potential hybridisation and introgression events mediated by this apparent anthropogenic connection.

Waipori-Tuapeka translocation

The dusky galaxias G. pullus is restricted to southeastern Otago, spanning adjacent headwaters of the neighbouring Taieri and Clutha systems (Figure 1A). Genome wide analysis (Figure 4B) reveals strong genetic differentiation between lower Clutha G. pullus samples relative to those from higher in the Clutha and Taieri systems (phylogeny in Figure 4B). Strong subcatchment structure is evident within lower Clutha samples (e.g. Waitahuna versus Tuapeka; Figure 4B). However, there is an anomalously close relationship between Tuapeka (Clutha) and Waipori (Taieri) samples that are separated by a major drainage divide, with upper Tuapeka samples (PT, PR) phylogenetically sister to a Waipori subclade (PB), and nested within the phylogenetic diversity of the Waipori as a whole (Figure 4B). Specifically, the Tuapeka samples and Waipori samples PB are phylogenetically linked, to the exclusion of the upper Waipori sample PP which has a sister relationship to this trans-divide clade. Based on the phylogeographic relationships recovered, it seems most likely that Waipori fish have colonised the upper Tuapeka. Regardless of the direction of gene flow, a clear mechanism for this recent and local trans-divide migration is provided by the anthropogenic water race connection between the Tuapeka and Waipori (Figure 3B). As above, the detection of recent trans-divide genomic connectivity in the precise location of an anthropogenic water race provides strong evidence for the role of drainage modification in redistributing genetic lineages.

Rough Ridge translocation

In Central Otago, two major catchments (Taieri; Clutha-Manuherikia) are separated by Rough Ridge (Figure 4C). In biological terms, the upper Taieri system is dominated by the flathead species G. depressiceps (Figure 1B; 4C), whereas the Manuherikia system (and much of the Clutha) is home to the widespread Clutha flathead lineage (G. ‘sp D’) (Figure 1B; 4C). Indeed, broadscale mtDNA data (Burridge et al. 2007; Waters et al. 2015; Craw et al. 2016a) indicate that Rough Ridge is generally a ‘barrier’ separating G. depressiceps and G. ‘sp D’ clade distributions (Figure 4C). In this context, the detection of G. ‘sp D’ genotypes in Totara Creek (within the Taieri range of G. depressiceps (Figure 4C)) represents a striking local biogeographic anomaly (see Figure 2). Specifically, intensive mtDNA and single-locus nuclear analyses of Rough Ridge Galaxias populations suggest recent genetic exchange across this divide (Figure 4C), likely mediated by a water race linking the headwaters of Pool Burn and Totara Creek (Figure 3C). This hypothesis is further supported by the fact that the water race itself currently provides habitat for Galaxias (Esa et al. 2000). This potential role of drainage modification as a mediator of hybridisation is supported also by recent genome-wide data (Campbell et al. 2022), with apparent hybrid genotypes detected in the water race itself, as well as in the nearby Manor Burn tributary of the Manuherikia (Clutha; Campbell et al. 2022). The presence of hybrid genotypes in the latter location suggests that G. depressiceps alleles may have introgressed widely into populations of the Manuherikia system. Future intensive genome-wide sampling on both sides of this Rough Ridge divide is required to more fully assess the regional extent of anthropogenic introgression, and the associated implications for conservation of the now fragmented fish populations occupying these systems.

Discussion

Artificial range shifts can have major implications for the evolution and conservation of freshwater-limited taxa (McDowall 1990), with hybridisation a frequent outcome (e.g. Echelle and Echelle 1997; Blackwell et al. 2021). Previous genetic analyses have highlighted the potential impacts of man-made drainage alterations (e.g. hydroelectric water connections) on fish distributions and diversity (e.g. Waters et al. 2002). Our review here synthesises evidence for anthropogenic range-shifts of fishes in three distinct locations, each apparently linked to past gold mining.

Based on the anthropogenic range-shifts identified here, we suggest that the penetration of foreign alleles into native populations may vary among systems, perhaps depending on the existence of ecological and physical barriers within and among streams (e.g. presence of predators and/or habitat breaks). For instance, Esa et al. (2000) detected no genetic signatures of ‘Taieri’ G. depressiceps introgression (Figure 4C) into G. ‘sp D’ beyond the Pool Burn region immediately west of the Rough Ridge water race. By contrast, recent genome-wide analysis of nearby Manor Burn G. ‘sp D’ samples (Campbell et al. 2022) hint that G. depressiceps genes may have introgressed more widely throughout the greater Pool Burn region. These new data highlight the potential of genomic approaches for revealing the geographic extent and impacts of such translocation events. On the eastern side of the Rough Ridge divide, spread of G. ‘sp D’ genotypes into the Taieri system has been detected in both Totara and Waimonga creeks, immediately adjacent to anthropogenic water race connections (Figure 4C; Esa et al. 2000), but not in the geographically intermediate Linn Burn (Figure 4C). However, future genome-wide approaches are needed to reveal the full geographic and genomic extent of anthropogenic admixture in this system.

In two of the three above cases of apparent anthropogenic fish translocation (Figure 4), the overriding genetic signature appears to involve dispersal primarily against the water-race flow (Figure 3). In addition to upstream dispersal of Pool Burn (Clutha) G. ‘sp D’ fish into the Totara (Taieri) system (Figure 4C; Esa et al. 2000), we here infer counter-current water-race dispersal of Waitahuna (Clutha) G. ‘Pomahaka’ into the Tokomairaro system (Figure 4A), and upstream movement of Waipori (Taieri) G. pullus genes into the Tuapeka (Clutha) system (Figure 4B). Although data are limited, these possibly asymmetric dispersal dynamics may have simple biological explanations. Specifically, strong-swimming larvae of the G. vulgaris complex (including those of G. pullus, and to a lesser extent G. depressiceps) typically exhibit positive (counter-current) rheotaxis from soon after hatching (Jones and Closs 2016a), with some taxa (e.g. G. pullus) showing little or no evidence of downstream larval dispersal (Jones and Closs 2016b). Members of this species complex also exhibit considerable climbing ability, and strong capacity for upstream dispersal (Allibone and Townsend 1997). Together, this combination of larval and adult behaviours may explain the inferred dispersal of Galaxias lineages against the prevailing flow of anthropogenic stream channels.

We have here presented evidence for genetic exchange events involving water race connections across three major drainage divides (Figures 3 and 4). The possible parallel effects of an additional four such inter-drainage connections (see Figure 1C) have yet to be assessed in detail due to a lack of fine-scale local sampling in the relevant regions. We predict that further sampling and analysis across these various regions of Otago will likely uncover additional evidence for trans-divide translocations (e.g. between Nevis-Nokomai; and between Waipori and upper Waitahuna). In cases where freshwater lineages have particularly narrow geographic distributions (e.g. Figure 1), there is potential for such translocations (with associated introgression/competition) to compromise both the genetic integrity and conservation status/value of affected taxa (Grabenstein and Taylor 2018; Blackwell et al. 2021). Alternatively (and perhaps counterintuitively), given that translocations can be an important tool for freshwater conservation management (Chilcott et al. 2013), it is possible that the range expansion opportunities afforded by anthropogenic translocation might improve the conservation status of some particularly range-limited or fragmented populations/taxa.

Broadly, ancient processes affecting river drainage connectivity are thought to have played important roles underpinning the divergence of freshwater-limited lineages in many regions of the globe (e.g. Kozak et al. 2006; Goodier et al. 2011; Craw et al. 2016b). In several cases, the persistent genetic legacies of such palaeodrainage features have been highlighted by interdisciplinary studies combining geological and biological data (e.g. Waters et al. 2020). The current study, by contrast, implies that recent (anthropogenic) drainage alterations in many locations have eroded and muddied the ancient phylogeographic signatures of geological history (Waters et al. 2015; Craw et al. 2016a). Specifically, the current synthesis highlights the role of anthropogenic drainage alteration in driving range-expansion of otherwise isolated and non-dispersive native fish lineages (Waters et al. 2020).

In broad terms, the examples explored here potentially represent tightly-constrained (both temporally and spatially) model systems for testing hypotheses regarding the conservation and evolutionary genetic effects of admixture. On the one hand, these recent headwater connections could be viewed as modern parallels to ancient, geologically-driven capture events which played key roles in freshwater biodiversification (e.g. Waters et al. 2001, 2015, 2020). On the other hand, such translocations can lead to competition, hybridisation and introgression, which can present major threats to species conservation (Rhymer and Simberloff 1996; Grabenstein and Taylor 2018). Moving forward, genome-wide approaches are clearly crucial for tracking the impacts of anthropogenic translocation events, and for prioritising, managing and conserving affected freshwater biodiversity (Dunn et al. 2018).

Acknowledgements

Tania King, Ludovic Dutoit, Graham Wallis, Yuzine bin Esa, and Richard Allibone provided important contributions to the concepts/data synthesised here. We thank Peter Unmack and Albert Chakona for their insightful comments which substantially improved the MS. Conceptualisation: JMW and DC. Biological synthesis and figures: JMW, CSMC. Geological synthesis and figures: DC. Writing: JMW, CSMC, DC.

Disclosure statement

No potential conflict of interest was reported by the author(s).