Historical translocations by Māori may explain the distribution and genetic structure of a threatened surf clam in Aotearoa (New Zealand)

The population genetic structure of toheroa (Paphies ventricosa), an Aotearoa (New Zealand) endemic surf clam, was assessed to determine levels of inter-population connectivity and test hypotheses regarding life history, habitat distribution and connectivity in coastal vs. estuarine taxa. Ninety-eight toheroa from populations across the length of New Zealand were sequenced for the mitochondrial cytochrome c oxidase I gene with analyses suggesting a population genetic structure unique among New Zealand marine invertebrates. Toheroa genetic diversity was high in Te Ika-a Māui (the North Island of New Zealand) but completely lacking in the south of Te Waipounamu (the South Island), an indication of recent isolation. Changes in habitat availability, long distance dispersal events or translocation of toheroa to southern New Zealand by Māori could explain the observed geographic distribution of toheroa and their genetic diversity. Given that early-Māori and their ancestors, were adept at food cultivation and relocation, the toheroa translocation hypothesis is plausible and may explain the disjointed modern distribution of this species. Translocation would also explain the limited success in restoring what may in some cases be ecologically isolated populations located outside their natural distributions and preferred niches.

limited inter-population connectivity is the toheroa (Paphies ventricosa) 14 , a large intertidal surf clam endemic to Aotearoa (New Zealand). Toheroa are broadcast spawners with a pelagic larval duration of around three to six weeks 15,16 . Toheroa inhabit exposed open-coast surf beaches and are primarily found in the middle of the intertidal zone. Juveniles are located at the upper end of this range (nearer the top of the beach) with adults located lower down the shore and buried up to 15-20 cm beneath the beach surface. The exact parameters that determine habitat suitability are uncertain (J. Cope unpublished research). However, in contrast to other New Zealand bivalves, such as the surf clam Paphies subtriangulata (tuatua) or the estuarine clam Paphies australis (pipi), toheroa appear to have habitat requirements that limit their geographic distribution 17 . Common features of the beaches on which toheroa occur include high wave energy conditions, a wide shallow gradient (dissipative beach) usually backed by sand dunes or cliffs, fine uniform sand with an average grain size of 0.21-0.33 mm, high levels of fresh water seepage onto the beach and high concentrations of phytoplankton 18,19 .
At the start of the 20th century, extensive toheroa populations were present on a handful of exposed west and south facing surf beaches (Fig. 1) 14 . Toheroa were a staple food for Māori (the indigenous people of New Zealand) in these areas and remain a species of great cultural importance. Toheroa began to be harvested more extensively by Pākehā (New Zealanders of European descent) from the late 1800s and intense harvesting over the next 60-70 years resulted in toheroa populations declining to levels where their harvest was no longer viable. Commercial and recreational fisheries were closed between 1969 and 1980 14 . Since that time only limited harvest for Māori customary purposes have been permitted. Despite having been protected for over 40 years, toheroa populations nationwide have, for unknown reasons, failed to recover, with some populations continuing to decline 14 .
The distribution of toheroa is unlike that of any other New Zealand marine invertebrate 2,14,17 (Fig. 1). Toheroa are largely absent from New Zealand's east coast and have a disjunct distribution over more than 1700 km (>11° latitude) of the western and southern coastline (Fig. 1). Notable populations have only been recorded in three regions. In Te Ika-a Māui (the North Island of New Zealand), toheroa can be found on the northland (Taitokerau) west coast from Muriwai to Te Oneroa-a-Tōhē (Ninety Mile Beach) and on the south-west (Kāpiti-Horowhenua) coast from Ōtaki to Foxton. In Te Waipounamu (the South Island) toheroa are largely restricted to Oreti Beach and Te Waewae Bay in Murihiku (the southern tip of the South Island) although there are accounts of small and transient populations in other southern locations. Numerous hypotheses have been proposed to explain the continued decline of toheroa including recruitment limitation and a lack of connectivity between geographically isolated populations 14 . Given the relatively short pelagic larval phase, and our knowledge of coastal circulation [20][21][22] , the exchange of toheroa larvae between northern and southern locations seems highly unlikely. Furthermore, a genetic boundary through central New Zealand is almost ubiquitous among marine invertebrates 2,23 , indicating possible barriers to north-south larval exchange.
In this study we assessed population genetic structure and levels of connectivity among toheroa across their entire distribution using the mitochondrial cytochrome c oxidase I (COI) gene. Our predictions, based on prior ecological knowledge, were for North Island populations to be well connected and for isolation and genetic divergence between northern and southern toheroa. Connectivity was expected to be less than that previously recorded for widely distributed open-coast taxa, such as P. sutriangulata, and more similar to levels of connectivity reported in studies of estuarine species such as P. australis 10 .

Results
Genetic diversity. Toheroa from Te Oneroa-a-Tōhē (NMB, n = 16) and Ripiro Beach (n = 16) in Northland, Waitarere Beach (n = 30) on the Kāpiti-Horowhenua coast and Oreti Beach (n = 36) in Southland were sequenced generating a 485 nucleotide COI alignment. Twenty-three nucleotide positions were variable leading to the delineation of 20 haplotypes (Table 1). Haplotype and nucleotide diversity was comparable across North Island populations with 12 haplotypes recorded for NMB/Ripiro (hereafter referred to together as Taitokerau; H D = 0.65, π = 0.0038) and 10 at Waitarere (H D = 0.56, π = 0.0036). In contrast, only a single haplotype (H1) was recorded at Oreti (H D = 0, π = 0). Haplotype H1 was also the most abundant haplotype across North Island populations, accounting for 59% and 66% of sequences from Taitokerau and Waitarere (Figs 2 and 3); Table 1). Of the remaining 19 haplotypes only one was shared among populations. Haplotype H2 was recorded at NMB (n = 1), Ripiro (n = 2) and Waitarere (n = 1). The remaining 18 haplotypes were private (unique to one location) with all but one of these recorded in just one specimen. A rarefaction curve generated from haplotype frequencies suggested that greater sampling effort could yield many more haplotypes, a notion supported by the number of missing intermediate positions indicated in a haplotype network (Fig. 2). Two groups of haplotypes were evident in this network (Fig. 2). Group 1, a star shaped haplogroup, contained 13 haplotypes separated from H1 by a single mutation and another three singleton haplotypes separated from H1 by one or two missing haplotypes. A second star shaped haplogroup was separated from the H1 haplotype by four to six mutations. This second group consisted of H2 and three additional haplotypes private to NMB and Waitarere.
Population genetic structure. AMOVA revealed statistically significant pairwise differences between all populations (p < 0.001; F ST = 0.39-0.74; Table 2) with differentiation between Oreti and Taitokerau toheroa (NMB and Ripiro) being greatest (F ST = 0.69-0.74; Table 2). Tajima's D and Fu's Fs were significantly negative for Waitarere (D = −1.692, p = 0.035; F = −3.815, p = 0.017) and for the COI data set as a whole (D = −2.187, p < 0.001; F = −17.639, p < 0.001; Table 1). This indicates evolution under non-random processes which could include directional selection and expansion or contraction. D and Fs could not be scored for Murihiku on account of the total lack of genetic diversity recorded at this site.

Discussion
This study provides support for the hypothesis that southern toheroa are isolated from toheroa in the North Island. However, the lack of genetic divergence between northern and southern populations is inconsistent with their distribution and dispersal capacity. Our results are also inconsistent with the expectation of connectivity among North Island toheroa, potentially indicating a dispersal capacity that is less than would be expected based on toheroa life history. The suggestion of expanding or contracting populations is consistent with toheroa dynamics over the last century as survey data indicate large fluctuations in toheroa abundance from year to year 24,25 . This dynamic was particularly evident in the decades leading up to the collapse and closure of the fishery, and is also relatively common in surf clams worldwide on account of their highly variable rates of recruitment success 26 . Oreti and Ripiro populations have tended to be more stable in recent decades 27 , while toheroa have now largely disappeared from the Kāpiti-Horowhenua coast 28 . This pattern of continued toheroa decline in Kāpiti-Horowhenua may account for the negative Fs and D values recorded for Waitarere, a population that may have limited larval connectivity with more northern toheroa.
Only two of the 20 haplotypes recorded in this study were found in more than one population and our analyses indicate statistically significant genetic differences between all regions. Genetic differences between North and South Island toheroa were the result of a complete lack of genetic diversity at Oreti in the South Island, a stark contrast to the relatively high levels of diversity recorded at each sampling location in the North Island ( Fig. 3; Table 1). Divergences between northern and southern populations have previously been reported for New Zealand marine invertebrates 2 including gastropods 29,30 , bivalves 10,31,32 , amphipod crustaceans 33 and echinoderms 34 . However, divergence typically takes the form of regionally unique haplotypes or geographical differences in allele frequency. These divergences are often indicative of an independent evolutionary history and reflect the limited connectivity between northern and southern regions 35 . In contrast to these previous taxa, the single haplotype recorded for toheroa at Oreti in the South Island was the same haplotype most commonly recorded in the North Island. This is a genetic structure that has not previously been reported for any other New Zealand marine invertebrates, even those with similar reproductive modes and pelagic larval durations 2,23 . For example, an analysis of microsatellite loci for P. subtriangulata 10 , a surf clam that can co-occur with toheroa, revealed one largely undifferentiated population across its entire distribution (North Island and upper South Island). In contrast, population subdivision (including divergence between northern and southern populations) was evident in the genetic structure of estuarine P. australis 10 (microsatellites) and Austrovenus stutchburyi 32 (mtDNA COI) with multiple alleles recorded in all populations. For A. stutchburyi, a number of COI haplotypes were found throughout New Zealand, but at different frequencies in northern and southern populations, while other haplotypes were found exclusively in either the north or the south.
A lack of within-population genetic diversity, as we observed for Oreti toheroa, could result from isolation and inbreeding 36,37 which is expected for the southern toheroa on account of their remoteness to North Island populations. If isolation had occurred in the distant evolutionary past we would expect some degree of divergence to have resulted after a period of isolation, and be manifested in the form of a set of unique genetic variants. Molecular clock estimates for marine molluscs suggest an evolutionary rate of between 2.3 and 4.6% per million years for the COI gene 38,39 . For the toheroa sequenced here, the six substitution steps between the central H1 haplotype and those at the farthest extent of the haplotype network could therefore represent somewhere between 270,000 and 539,000 years of toheroa evolution in northern New Zealand. In contrast, the lack of divergence or genetic diversity in southern toheroa suggests a far shorter evolutionary history and supports the notion of a more recent isolating event with no subsequent opportunity for divergence. Based on the above rates of evolution, divergence in Oreti toheroa from an ancestral population should be detectable after c. 45,000-91,000 years of isolation. Accordingly, the absence of unique haplotypes in the south suggests these populations were founded more recently.
Three scenarios could explain the geographic distribution of genetic diversity recorded in toheroa: First, that toheroa were formerly present along east or west coasts of the South Island (connecting northern and southern populations) and have disappeared from these areas due to factors such as over harvesting, the loss of suitable habitat or a changing climate. The occurrence of toheroa in Māori shell middens (domestic refuse heaps that indicate sites of human occupation or food processing activity) in the upper and central South Island would support this hypothesis if a range reduction had occurred after the settlement of New Zealand by Māori in approximately 1250 AD 40,41 . However, while toheroa shells are found both in middens and in natural deposits in the vicinity of present day toheroa populations (New Zealand Archaeological Site Recording Scheme), they have rarely been reported elsewhere. There is no archaeological evidence for toheroa populations at other locations and the habitat along east and west coasts of the South Island is, at present, largely unsuitable for toheroa (P. Ross pers. obs.).
A second possible explanation is that rare, long distance larval dispersal events from northern populations may have seeded toheroa in the south. While it is impossible to rule out this scenario, our understanding of coastal circulation patterns suggests that the delivery of toheroa larvae over distances of 800 km or more is unlikely 20,22 . However, if a founding dispersal event had occurred in the distant past (>45,000 to 91,000 years ago), genetic divergence from northern toheroa would be expected 42 . Alternatively, a more recent dispersal and founding event could have occurred, explaining the lack of divergence recorded in this study. However, there is currently no evidence to suggest that hydrodynamic conditions have changed such that long distance dispersal events would have only become possible in the recent past and this explanation seems unlikely.
A third possibility is that toheroa are not native to the South Island, but were translocated there by Māori. Early-Māori were prolific users of aquatic resources 43,44 and were adept at food cultivation and translocation. The Polynesian explorers who settled New Zealand (the first Māori) brought with them plants (sweet potato, taro and yam) 45,46 and animals (Pacific dogs and rats) 47 . Post settlement, Māori also domesticated and translocated numerous endemic plants 48 and are thought to have translocated freshwater fish into lakes where they did not occur naturally 38 . Not surprisingly, it has been argued that translocation and active management of wild populations are part of Māori culture [49][50][51] . Despite the overwhelming evidence that cultivation and translocation were commonly used by early-Māori, the concept of pre-European translocation of marine species has received little attention and no attempts have been made to determine whether practices similar to those employed on land were also used in the marine estate -in this case, the cultivation of shellfish through translocation. An examination of data sources typically outside of those utilised by practitioners of natural sciences (including legal and historical texts, popular media and interviews) has provided numerous accounts of toheroa translocation within both North and South Islands dating back at least as far as c. 1880 AD [52][53][54][55] . Based on this evidence, we now know that toheroa translocations took place. Translocations may even explain the small numbers of toheroa or their shells that have been reported at sites such as Pakiri or Karamea where toheroa are not found today (Fig. 1). However, at this time it is unknown whether the presence of toheroa in southern New Zealand, or on the east coast of the North Island, is solely a consequence of translocation. Further studies, incorporating archaeology, anthropology and molecular ecology (using more variable genetic markers), may address this uncertainty.
While the genetic structure of toheroa appears to be unique, there are few other relevant taxa for which similar data are available 2,23 . Population genetic analyses of New Zealand bivalves are limited to A. stutchburyi 32 , Paphies australis, Paphies subtriangulata 10 , and Perna canaliculus -a mussel with a population genetic structure that has to some extent been modified through modern aquaculture-related movements of broodstock and/or spat 33,56 . Consequently, we caution that other presently unstudied New Zealand bivalve taxa could have similar population genetic structures to that of toheroa and that these patterns may have arisen through natural biogeographic processes.
The reasons for the failure of toheroa to recover despite over 40 years of protection remain a mystery. The detection of a population genetic structure that is unique among New Zealand marine invertebrates 2 adds to the intrigue surrounding this iconic species. While limited gene flow is just one of many factors potentially influencing toheroa population dynamics 14,57 , our study provides support for the hypothesis that South Island toheroa are genetically isolated. Furthermore, the limited sharing of haplotypes among northern populations may indicate limited connectivity at smaller spatial scales. As genetic diversity in northern toheroa is high, analyses that capture a greater proportion of this diversity will be required to provide estimates of larval exchange and will allow for comparisons with coastal and estuarine taxa. Such comparisons, and future population genetic studies, should incorporate thinking around the potential for humans to have influenced the distribution of marine species. If the translocation hypothesis is accepted, it could explain the disjointed modern distribution and population dynamics of toheroa and the limited success achieved in restoring what may, in some cases, be ecologically isolated populations located outside their natural distribution and preferred ecological niche.

Methods
Study sites and sample collection. Specimens for genetic analysis were obtained from the three main toheroa regions (Fig. 1). In northern New Zealand, toheroa were sourced from NMB (n = 10) and Ripiro Beach Population genetic analysis. Indices of genetic diversity were quantified using DnaSP Version 5 59 and Arlequin Version 3.5.2.2 60 . For each population (Taitokerau (NMB and Ripiro), Waitarere and Oreti) we calculated the number of segregating sites (S), numbers of transitions (T S ) and transversions (T S ), the number of COI haplotypes (N H ), the number of private COI haplotypes in a population (H P ), haplotype diversity (H D ), mean number of pairwise differences (k) and nucleotide diversity (π). A rarefaction analysis was generated in Analytic Rarefaction (Version 1.3; http://strata.uga.edu/software/anRareReadme.html; Holland 2003) to approximate the proportion of haplotype diversity captured by the current sampling regime. PopART Version 1.7 (http://popart. otago.ac.nz) was then used to generate a minimum spanning network to allow for visualisation of relationships between haplotypes and the geographic distribution of genetic diversity. Estimates of population pairwise F ST values were then calculated in Arlequin to determine if any two populations differed significantly in their genetic composition. Tajima's D 61 and Fu's Fs 62 were calculated in DnaSP to test for deviation from the Wright-Fisher model of neutral evolution which can be indicative of either non-neutral evolution or population expansion or contraction under neutral evolution.