The spatio-temporal axis for phenotypic change: a comparison of source and translocated Arctic charr populations after 25 generations

Evolution of morphological traits is hypothesized to act on an extended time scale, yet studies have suggested that these changes are possible within a few generations. Trophic polymorphism enabled through niche adaptations and ecological opportunity is one phenomenon that facilitate occurrence of rapid adaptive variation, common in many northern freshwater fish species. One such species is Arctic charr, which is known for its extensive variation in morphology and the occurrence of morphs. However, the speed at which such morphological variation arises is poorly studied despite the importance for understanding the onset of evolution. The aim of this study was to elucidate this process in a gradient of eight lakes that was stocked with Arctic charr in the period from 1910 to 1917 from Lake Tinnsjøen, Norway. We used morphological measurements to test for differences in traits between populations and Haldane and Darwin’s evolutionary rates to estimate divergence rates in traits. We also tested for correlation between putative genetic and morphological divergence. In addition, we contrasted the morphological divergence with that expected under neutral genetic expectations, using 12 microsatellite markers, to analyze whether and which morphological differences that is following early genetic divergence. A significant genetic differentiation was found between the source population and five of the translocated populations with corresponding differences in morphological traits for four of the populations. Population genetic structuring indicated six different genetic clusters. The translocated populations also exhibited trait divergence estimated with both Haldane and Darwin’s rates. Differences in morphological traits showed a significant correlation with genetic divergence, and the morphological differences were most likely affected by differences in lake parameters such as maximum depth, lake size and fish community. We conclude that intraspecific morphological and genetic divergence can form on short evolutionary time scales with important implications for conservation and management practices.


54
Adaptive changes in phenotypic traits mediated by natural selection is a crucial component in 55 evolution resulting in population differentiation (1). In this process, adaptation to different 56 physical environments, such as divergent niches, temperature and climate regimes, as well as (Oncorhynchus nerka) which diversified into two reproductive isolated populations after 13 114 generations following introduction, resulting in significant differences in body shape (57).

116
In Lake Tinnsjøen, Norway, the Arctic charr have been documented to consist of four 117 morphologically and genetically distinct morphs; the planktivore, dwarf, piscivore-, and population to more or less replicated spatially separated environments 100 years ago, lending a 145 novel opportunity to study phenotypic and genetic changes. Thus, the aim of the study was to 146 investigate the morphological and genetic response as well as rates of differentiation by 147 studying translocations of Lake Tinnsjøen Arctic charr into neighboring, but different, lake 148 environments. Study systems 153 A hatchery existed in Lake Tinnsjøen producing offspring of wild caught Arctic charr parents 154 where the main intention was to stock Arctic charr in nearby lakes for human consumption. We 155 assume that the most common Arctic charr morph (which is also the most appreciated as a food 156 source) was used for breeding in the hatchery. This is the planktivorous morph (referred to as 157 the "normal morph" or "pelagic charr"). More details of the diversity of the Lake Tinnsjøen  ; Table 1).

164
With regard to Lake Finsevatn, the introduction probably took place in connection with the 165 introduction of 30 000 charr in surrounding lakes at this time (68).  Gill-nets (Nordic multimesh gillnet series) (mesh size described in (58)) were set in all lake 199 habitats (littoral, pelagial and profundal). Benthic gillnets were placed from land in the littoral 200 zone, while profundal gill-nets were placed at the deepest point in each lake, and floating 201 gillnets were placed in the open water at depths between 0-6 meters. In accordance with lake 202 size, depth, species composition, and local knowledge, catch effort was adjusted among 203 localities from a baseline of four littoral-and profundal nets, and one pelagic net. In a medium 204 sized lake such as LU, we increased the effort in the pelagic area to three gillnets, while in the 205 smallest lake, TJ, only four nets were set in total. All nets were set to fish overnight for 12-206 hours.

208
After catch, the right pectoral fin from all Arctic charr was stored on 95% ethanol for DNA 209 analysis. Fish were stored in an ice cooler and brought to a freezer (-40 °C) for storage. The  From all lakes (except FI) we measured transparency and water color using a secchi disc, while 213 maximum depth was recorded using a handheld echo sounder (Plastimo Echotest II). Water  Unfortunately, no sampling of invertebrate or zooplankton was performed in Lake FI. was calculated for all landmarks to describe differences in individual variation in body shape.

234
To correct for length effects a regression between shape data and log centroid size was   difference, models were considered as equal.

252
A set of eight metric and two meristic traits were analyzed as these traits may reflect phenotypic   The divergence rates for all the ten phenotypic traits between the source population of Lake  outgroups from southern and northern Norway was performed to evaluate clustering (Fig. S4).

376
Redundancy analysis (RDA) was performed to assess how much of the genetic variation 377 reflected the observed phenotypic variation, and to detect the main traits that associated with 378 the genotypic variation. In order to prevent overfitting, the command "OrdiStep" in the R 379 package VEGAN, was used for forward selection of morphological traits using residuals (97).

380
Overall F ST was calculated with the Hierfstat R package (98). As the total variation of a PCA   clustered into one single group with regard to body shape (Fig. 3 & S3). Arctic charr exhibited significant differences, with Lake SN having the longest lamella (mean 429 7.2), and Lake GR exhibiting the shortest lamella (mean 3.2). Head length (LH) differed 430 significantly for Arctic charr from Lake FI, GR, SV, SØ, TJ and TO. Arctic charr from Lake 431 TJ exhibited the longest head length (mean 51.9), and Arctic charr from Lake GR had the 432 shortest head length (mean 27.5). Eye area size (EA) differed significantly between Arctic charr 433 from Lake GR and SV, where Arctic charr from Lake SV had the largest eye area (mean 44.5) 434 and Arctic charr from Lake GR had the smallest eye area (mean 27.6). Finally, length of the 435 pectoral fin differed significantly between Arctic charr from Lake GR, SV, SØ and TJ. Here,

436
Arctic charr from Lake TJ had the longest pectoral fin (mean 46.2) and Arctic charr from Lake 437 GR had the smallest pectoral fin (mean 23.6, Table 2).    Dimension 1 is along the X-axis and Dimension 2 is along the Y-axis (see Fig. 3).

459
Here populations TI, FI, LU and SN were clustered into one group, and populations GR, TO,  The MANOVA analysis showed that both maximum depth and size of the lake had a significant between lake size and morphological and meristic traits (   The rates of trait divergence from Lake Tinnsjøen to the translocated populations based on 506 Darwin's (kilodarwin's) and Haldane's ranged from -7.81 to 1.39, and from -0.48 to 0.20, 507 respectively ( Table 6). The three populations that exhibited the highest rates of divergence with the source population (Table 6).

533
Significant genetic differentiation (F ST ) was found between most of the Arctic charr from the 534 eight translocated populations and Lake Tinnsjøen Arctic charr. The exception was non-535 significant differentiation between Arctic charr from Lake TI and FI, TI and LU, TI and SØ, 536 and FI and LU (Table 7). F ST´s ranged from 0.01 between Arctic charr from Lake FI and TJ, to 537 0.14 between Arctic charr from SN and TO. The population pairs exhibiting the highest F ST 538 values (above 0.11) were Arctic charr from; Lake TI and GR, SN and SV, TO and TI, SØ and 539 TI, and between SV and TO and TI (Table 7).   Testing for population genetic structure using STRUCTURE, a hierarchical approach was 550 performed where the initial run suggested 6 genetic clusters (deltaK = 5.623, mean LnP(K) = 551 7459.9, Fig. S2). Here, further runs were performed where the most deviating population was 552 removed from each run to investigate population structure. The most likely partition was thus 553 six clusters, where no significant genetic differentiation was observed between Lake TI and 554 three of the translocated populations (FI, LU, and SØ) (Fig. 5). The phylogenetic neighbor-joining tree suggested a partition into three main branches where 562 Lake TI and FI resided on the main branch, whilst only a minor bootstrap support of 23% 563 separated Lake GR, SN and SV populations LU, TO, SØ and TJ into the next two main 564 branches. Minor bootstrap support (18%) separated Lake GR from Lake SN and SV, while low 565 support (49%) separated Lake SN from Lake SV. Low bootstrap support of 50% separated Lake 566 LU from Lake TO. Likewise, a low bootstrap value of 45 % separated Lake TO from SØ and 567 TJ, while Lake SØ and TJ were separated by a moderate bootstrap support of 67% (Fig. 5). The 568 second neighbor-joining tree including five outgroups showed a similar pattern where the 569 outgroups resided in a separate branch from the remainder of the populations (Fig. S4). The overall F ST value calculated from the genetic data was 0.072. The first 25 PCA axes that 574 explained 50% of the total variation (Fig. 6) was used to illustrate genetic variation in the 575 redundancy analysis. The morphological traits that best explained the genetic variation, i.e.,

576
where variation in phenotypic traits reflected genetic variation, were LG, PCA2, GR-L and L.

577
When analyzing all morphometric traits, 6.9% of the total variation in the genetic data was 578 associated with the phenotypic variation, and the four traits (LG, PCA2, GR-L and L) that 579 explained the data best accounted for 3.8% (Fig. 7). With overall F ST of 0.072, the 6.9% rates was the number of gill rakers on the lower arch (GR-L). Similar results in higher than expected means of Haldane rates was found in a study on morphological traits by Michaud,642 Power (56) in a translocated population of Arctic charr after only 25 years of isolation, which 643 is more rapid than several documented natural cases (106). It is possible that the charr in most 644 of the translocated lakes exhibits a decreased number of gill-rakers as a response to a decreased 645 pelagic area (and in some cases also a lack of interspecific competition) compared to Lake TI, 646 and thus potentially a decreased dependency on zooplankton as the main prey group.  source where the parents were taken from. There was also a correlation between the genetic 706 and morphological changes in the translocated Arctic charr. However, it seems likely that for 707 these changes to occur, the environment in which the Arctic charr is translocated to, need to 708 inhabit certain characteristics, such as high niche heterogeneity in the form of large, and/or deep 709 lakes and with a species composition that enables divergence to different niches (16, 26, 64).

710
One important aspect of this study is that if the purpose is to study rapid divergence and