A behavioral syndrome linking boldness and flexibility facilitates invasion success in sticklebacks

For a species to expand its range, it needs to be good at dispersing and also capable of exploiting resources and adapting to different environments. Therefore, behavioral and cognitive traits could play key roles in facilitating invasion success. Here, we show that dispersing sticklebacks are bold, while sticklebacks that have recently established in a new region are flexible. Moreover, boldness and flexibility are negatively correlated with one another at the individual, family and population levels. Multiple lines of evidence suggest that the divergence in boldness and flexibility is likely to be evolutionary in origin. If boldness is favored in invaders during the initial dispersal stage, while flexibility is favored in recent immigrants during the establishment stage, then the link between boldness and flexibility could generate positive correlations between successes during both dispersal and establishment, and therefore play a key role in facilitating colonization success in sticklebacks and other organisms.

Understanding the factors that allow a species to expand its range and adapt to changing 56 habitats is increasingly important in the face of anthropogenic change. Natural biological 57 invasions can reveal how and why certain organisms can excel in response to novel selection 58 pressures (Whitney & Gabler 2008). In addition to the importance of propagule pressure 59 (Lockwood et al. 2005), stochasticity, and the opening of new niches at the edge of species 60 boundaries, there is growing evidence that particular traits (e.g., r-selected life histories 61 (Capellini et al. 2015), habitat breadth (Blackburn et al. 2009), large brains/cognitive abilities 62 (Sol et al. 2005 Additionally, previously successful behavioral patterns may no longer be successful in new 75 environments, so immigrants need to be able to stop persisting on ineffective responses and 76 flexible enough to attempt new approaches (Griffin et al. 2016). 77 Successfully colonizing a new environment can be broken down into discrete stages, e.g., 78 dispersal, colonization, establishment and spread (Chapple et al. 2012), and invasion success 79 likely relies on different behavioral and cognitive traits in each stage (Sih et al. 2012 Threespined sticklebacks (Gasterosteus aculeatus) are a model system for studying trait 99 evolution during biological invasions. Throughout their evolutionary history, marine sticklebacks 100 have repeatedly colonized freshwater environments, rapidly adapted to them and diversified 101 (Stuart et al.). Sticklebacks can also spread and have dramatic impacts on freshwater 102 communities (Eklöf et al. 2020;Gugele et al. 2020). Work on this system has primarily focused 103 on a suite of morphological and physiological traits that repeatedly evolves when marine 104 sticklebacks invade freshwater habitats (Bell et al. 2004;Colosimo et al. 2005), with evidence 105 that haplotypes containing a set of coadapted alleles are maintained at low frequency in the 106 ocean and are repeatedly tapped during adaptation to freshwater (Bassham et al. 2018). 107 However, little is known about the behavioral and cognitive mechanisms that may facilitate the 108 invasion of sticklebacks into new habitats and which may evolve once a population becomes 109 established (Foster et al. 2015). Eggs were fertilized in the field following previously established protocols (see (Wund et 138 al. 2012)). Two to three days post fertilization, the eggs were transferred to 50 mL canonical 139 tubes and shipped overnight in coolers filled with ice packs to the University of Illinois Urbana-140 Champaign where they were raised in common garden conditions in the lab. Artificial 141 incubation controls for environmental paternal effects due to receiving paternal care therefore it 142 is likely that phenotypic differences among the lab-reared populations reflect heritable 143 differences (although environmental maternal effects could also contribute). 144 Clutches were reared in separate tanks (9.5L 32 x 21 x 19 cm) where the embryos were 145 incubated in a cup with a mesh bottom and placed over an air bubbler. Fish were kept at 60°F 146 with an even light cycle (12L:12D) for the entirety of the experiment. All families were kept on 147 one of two recirculating flow-through water racks, which consisted of a series of particular, 148 biological, and IV filters and had three different shelves (Aquaneering , San Diego, USA). 10% 149 of each tank's water was replaced each day. Family tank position was pseudo-randomly assigned 150 so that all populations were evenly distributed across both racks and the three levels of shelves. 151 Importantly, we elected to rear the fish and measure their behavior in freshwater (~5ppt), thereby 152 simulating the conditions that marine sticklebacks encounter when they move into freshwater. In order to ensure that an individual had acclimated to their home tank and was motivated 183 to eat during the behavioral tests, the individual was presented with food via a petri dish at the 184 center of their home tank, and the individual had to eat the food within 10 minutes on three 185 consecutive days in order to proceed to the next step. On average, it took 5.1 days for fish to 186 meet this criterion (range = 3 to 17 days). 187 188

Novel object test (neophilia) 189
Individuals' response to a novel object (toy lion; 10L x 7H cm; TERRA by Battat,190 Montreal, Canada) was recorded the day after the fish met criterion in the acclimation phase. The 191 toy lion was selected as a novel object because the fish had no prior experience with this object, 192 there was no presumed evolutionary history with the object's shape, and it was made up of 193 neutral colors. 194 The individual's behavior was measured for five minutes after their first approach to the 195 novel object (i.e., first time within one body length of the novel object and oriented directly 196 towards it). We interpret more time spent near and oriented towards the novel object as greater 197 time investigating the object (i.e., higher neophilia). 198 199

Latency to emerge (boldness) 200
For each individual, we recorded their latency to emerge from a refuge on their first four 201 training trials for the barrier task (described below) and used the average of those measures as a 202 proxy for boldness (Movie S1). Latency to emerge was repeatable across the four trials within 203 each population (r=0.56-0.72, Table S4). 204 205

Barrier detour task (flexibility) 206
Pretraining for the barrier detour task started on the same day the novel object test was 207 completed. The goals of pretraining were to train the individual to learn that there would 208 consistently be a food reward in the middle of the tank, establish food motivation in this context, 209 and create a prepotent response of leaving a shelter to directly approach and eat the food reward. 210 During pretraining, individuals were trained for one session per day, with each session 211 comprising four trials. To start the trial, the observer removed a cork from the side of the shelter 212 and the fish was given ten minutes to exit. Upon exiting the fish was allowed five minutes to eat 213 the worm. 214 After eating the worm, the fish was placed back into the shelter in preparation for the 215 next trial. If the fish did not emerge from the shelter within ten minutes after the cork was pulled 216 or eat within five minutes after emergence, the observer recorded the maximum times for these 217 behaviors, removed the food reward and gently poured the fish out of the shelter if necessary. 218 Latency to emerge from a refuge during the first pretraining session was used as a measure of 219 boldness (see above). 220 Training for the barrier task was criterion based. In order to move on to the barrier task 221 following pretraining, the individual had to emerge from the shelter within 10 minutes and 222 directly approach and eat the food reward within five seconds on three out of the four trials. The 223 one failed attempt could not be on the fourth trial; this requirement ensured that the fish would 224 be motivated throughout the four trials. Fish were given a maximum of four days to reach 225 criterion. Fifteen fish did not meet criterion (Big Beaver: 7, Cornelius: 1, Loberg: 1, Cheney: 3, 226 Rabbit Slough: 2, Resurrection Bay: 1). 227 Once an individual met criterion, the individual moved on to the barrier detour task the 228 following day. This task also comprised four trials. The first two trials were exactly the same as 229 the pretraining trials in order to reinforce the direct search pattern. On the third trial a transparent 230 semi-circular barrier was placed between the shelter and food reward. The opening into the 231 barrier area was positioned directly in front of the entrance to the shelter. After removing the 232 cork the individual was allowed 30 minutes to emerge from the shelter, navigate around the 233 barrier and eat the food reward (Movie S2). The observer recorded the duration of the first bout 234 (no break in contract longer than five seconds) at the apex of the barrier. In order to confirm that 235 the fish that spent little time at the barrier apex during the third trial were still motivated to eat, 236 the fish's behavior was observed for a fourth trial during which no barrier was present.  (Table S2). 305

Neophilia did not evolve systematically during the invasion 307
Neophilia (time near and oriented to a novel object) did not vary among populations or 308 between the sexes (Table S1) and did not vary in a systematic manner among the different types 309 of populations (Figure 3). Larger fish were less neophilic (β = -0.033, t = -2.631, df = 247.79, 310 p=0.009, Figure S2, Table S1). We did not detect variation among families within populations in 311 neophilia, as FamilyID did not significantly improve model fit (Χ 2 = 2.369, df = 1, p = 0.124; 312 AIC with = 636.57; AIC without = 636.94). Estimates of broad sense heritability of neophilia in the 313 six populations ranged from 0.09 to 0.87 (average = 0.34) and was significantly different from 314 zero in the Cornelius population (Table S2). 315

A boldness-flexibility syndrome facilitates invasion success 317
Given that sticklebacks from the dispersing populations were bolder and less flexible 318 compared to sticklebacks from established populations, we tested how boldness and flexibility 319 were correlated with one another. Consistent with the pattern described above, individuals that 320 were more bold (quickly emerged from the refuge) were less flexible (spent more time at the 321 barrier apex; r = -0.522, n = 245, p<0.001, Table S3), and this pattern was also evident at the 322 family (r = -0.688, n = 62, p<0.001) and population (r = -0.91, n = 6, p=0.01) levels (Figure 4). 323 The average genetic correlation between boldness and flexibility within the six populations was r 324 = -0.69 (range: -0.59 to -0.77) and was significantly different from zero in the Cornelius 325 population (Table S3). 326

DISCUSSION 328
The possibility that behavioral and cognitive traits might facilitate and evolve during 329 natural biological invasions is intriguing but difficult to study directly. Here, we took advantage 330 of a model system for invasions to test the hypothesis that boldness is favored in dispersers, that 331 neophilia and flexibility are favored in recently-arrived immigrants and that these traits are 332 subject to relaxed selection and possibly lost once a population becomes established in a new 333 environment. We found that boldness and flexibility evolve in a systematic way when marine 334 sticklebacks colonize freshwater habitats. Specifically, the dispersing populations were bold, and populations (Cheney Lake and Loberg Lake, respectively) is also insightful; the two newly 364 derived populations either tended to resemble the dispersing populations or the well-established 365 populations. Cheney Lake was founded more recently than Loberg Lake (9 versus 28-34 years 366 prior to this study); if it takes longer than 10 generations for behavioral and cognitive traits to 367 diverge from the ancestral marine behavioral type then time since establishment could be 368 important. Alternatively, or in addition, the phenotypic differences between sticklebacks from 369 Loberg Lake and Cheney Lake could reflect differences in the way that the two lakes were 370 colonized; Loberg Lake was naturally colonized, while sticklebacks were experimentally 371 introduced to Cheney Lake. If particularly flexible individuals were more likely to disperse into 372 Loberg Lake, but a random sample of behavioral types were artificially introduced into Cheney that bold individuals are more likely to disperse, and that boldness and flexibility are tightly 377 negatively correlated with one another, this explanation seems unlikely. Further studies tracking 378 how behavioral and cognitive traits change over time in the Cheney Lake population (and similar 379 experimental lakes (e.g., Scout Lake)) could help discriminate between the assortative mating, 380 time since establishment and non-random dispersal hypotheses. 381 We originally hypothesized that neophilia would be favored in recently-derived 382 populations because seeking and/or being willing to investigate novel stimuli may help newly-383 arrived immigrants locate new habitats and discover novel resources, but we found no support 384 for this hypothesis. One possible explanation for the failure to find systematic differences in 385 neophilia among the populations is that neophilia may actually be disadvantageous in a new 386 environment because it can expose animals to dangerous stimuli they have never encountered 387 before. Another potential (nonexclusive) explanation based on our results is that neophilia may 388 not evolve as readily because it may be less heritable (effect of FamilyID was nonsignificant, 389 lower H 2 estimate). Instead, neophilia may be more influenced by age or experience: in this 390 study, smaller (and younger: r = -0.159, t = -2.595, n = 262, p=0.01) fish were more neophilic, 391 which is consistent with other studies which have shown that novelty-seeking decreases with age 392 (Stansfield & Kirstein 2006). 393 Sticklebacks are a powerful model system for understanding how and why certain traits 394 repeatedly evolve whenever organisms invade new habitats. Accumulating evidence suggests 395 that sticklebacks have evolved mechanisms for rapidly adapting to new environments with 396 alleles conferring the freshwater-adapted phenotype maintained at low frequency in the ocean 397 (Colosimo et al. 2005). But the role of behavior and cognition in facilitating evolutionary 398 processes in this system has received less attention (Foster et al. 2015). Our results suggest that a 399 behavioral mechanism -a behavioral syndrome linking boldness and flexibility together -can 400 contribute to rapid adaptation in this and other organisms.