Evolution of Daphnia population dynamics following invasion by a non-native predator

Predators are frequently observed to cause evolutionary responses in prey phenotypes, which may, in turn, translate into evolutionary shifts in prey population dynamics. Although a link between predation and population growth has been demonstrated in experimental evolution studies, insights from natural populations are lacking. Here we tested for evolutionary changes in the population dynamics of the herbivorous water flea Daphnia pulicaria in response to the invasion of the predatory spiny water flea (Bythotrephes longimanus) in the Great Lakes region, USA. Using a resurrection ecological approach and a 3-month population growth experiment (in the absence of predation) we compared population dynamics in daphnia from pre- and post-invasion time periods. Post-invasion daphnia were able to maintain an overall higher population abundance throughout the growth experiment, both in terms of the number of individuals (28% higher) and total population biomass (33% higher). Estimation of population dynamics parameters from a theta-logistic model suggested that this was achieved through an increase in intrinsic population growth rate as well as increased carrying capacity. The observed difference in intrinsic rate of increase could not be predicted based on previous measurements of life-history traits in these clones. This indicates that care should be taken when extrapolating from a few life history traits measured in isolated individuals under controlled conditions to population dynamics. Whereas previous experimental evolution studies of predator-prey interactions have demonstrated that genotypes that have evolved under predation have inferior population growth when the predator is absent, this was not the case for the Daphnia. We suggest that complexities in ecological interactions of natural ecosystems, such as the potential for spatial and temporal avoidance of predation, makes it challenging to provide general predictions about evolutionary responses in population dynamics to predators.

available. An alternative approach is to quantify genetically based differences in life-history 79 traits (e.g. age at maturation, age-specific fecundity) in a common environment using individuals 80 originating from different predation regimes, and then use these to infer an intrinsic rate of 81 population growth (i.e. r, Gillis & Walsh, 2017). Critically though, such data are typically 82 obtained under ad libitum food availability and in the absence of density effects, meaning that a 83 'true' ecological context will be lacking in most cases.  Thompson, 1998).

89
The effect of predation has received particular attention within ecological studies of freshwater 90 zooplankton communities. In such communities, populations tend to fluctuate naturally in 91 abundance throughout the year due to feedbacks between of population density and food 92 availability (Sommer et al., 1986). Due to the importance of density and food abundance in        The transparent tray was put on a LED light board (Huion A4 LED light pad, set to maximum 172 intensity) in a dark room directly under the camera (Basler aCA1300-60gm, fitted with 5-50-173 mm, F1.4, CS mount lens) and filmed for 10-14 seconds. We waited for any movement of the 174 medium to stop before starting each recording. The contents of the tray were then emptied back 175 in the original plastic container. This procedure was repeated for each population. After 176 finishing, each container was filled to 100mL with ADaM and 1.5 mL shellfish diet was added.

177
The larger ration during this event than during regular feeding was done to compensate for the 178 complete removal of all algae and bacteria during medium change. The procedure described 179 above produced a total of 1657 videos that were subsequently analyzed to obtain data on 180 population sizes. The first 39 frames from each video were extracted as images using a custom   Inspection of population growth rate data (both for numerical and biomass growth) revealed 204 strong non-linearity in the density dependence. Thus, the population dynamics is best described 205 by the theta-logistic model G = r(1-(Nt/K) θ ) where r is the intrinsic population growth rate, K is 206 the carrying capacity, and θ determines the shape of the density dependence. One concern when 207 fitting this model to data is that different combinations of r and θ can produce model fits of  Next, to test for an effect of invasion history on K we fitted non-linear mixed effect models 235 representing the theta-logistic model to the population growth time series. This was done for the 236 data following the first two censuses, i.e. after the period that had been used to calculate r. 237 Again, this was done separately for numerical and biomass growth rates. In the full model, 238 population growth rate over a given period (between two consecutive censuses) was modelled as 239 a function of the observed value of r for that population and its population size at the start of that

249
Population size trends for both pre-and post-invasion populations showed a period of high 250 population growth up to a peak level, followed by a slower decline (Fig. 1). This was true both 251 for numerical abundance and biomass. For both measures of population size, post-invasion 252 clones maintained higher values than pre-invasion clones throughout the experimental period 253 (Fig. 1, Table 1). Model parameter estimates suggest that post-invasion clones attained an overall 254 increase relative to pre-invasive clones in numerical abundance by 28%, and in biomass 255 abundance by 33% (Table 2). Observed values of intrinsic population growth rate also tended to depend on invasion history 258 (Table 3). The evidence for such an effect was strongest for biomass (Table 3), but models 259 containing an effect of invasion history suggested higher intrinsic population growth rate in post-260 invasive clones than in pre-invasive clones for both numerical and biomass growth (Table 4).

261
The estimated increase in intrinsic population growth rate based on numerical and biomass data 262 were 23 and 15%, respectively (Table 4).

264
The difference in population dynamics between pre-and post-invasion clones was also reflected 265 in the comparisons of theta-logistic models of population growth rates. For both numerical and 266 biomass data, models containing an effect of invasion history on the carrying capacity received 267 more support than models without this term ( Table 5). The strength of evidence for such an 268 effect was particularly pronounced for numerical data. For both these analyses, the best models 269 predicted a higher carrying capacity for post-invasion clones compared to for the pre-invasion 270 clones (Table 6, Fig. 2). The estimated increase in carrying capacity based on numerical and 271 biomass data were 27 and 25%, respectively (Table 6).  showed that algae that had evolved under predation exposure had a lower population growth rate 311 than those having evolved in absence of predation, but only at the most limiting nutrient-level.

312
When nutrients were more abundant no such difference was observed. Thus, the authors 313 conclude that the algae face a trade-off between competitive ability and predator resistance, and 314 additional lines of evidence suggest that this is due to selective grazing by rotifers on larger cells,  In the current study we found no evidence for an interaction between predation history and food 318 availability in determining the rate of population growth as observed in the rotifer/algae 319 experiments. Rather, clones that had evolved in the presence of the predator were able to 320 maintain an overall larger population abundance throughout the experiment (and thus in response 321 to declining levels of food as population density increased) in the absence of the predator. 322 Furthermore, models that contained an effect of invasion history consistently suggested elevated 323 values of both r and K in post-invasion clones compared to pre-invasion clones, independent of 324 measurement type (numerical or biomass population dynamics). Thus, our results also fail to 325 provide support for an evolutionary trade-off between the intrinsic rate of increase and strength 326 of density dependence that has been predicted on theoretical grounds (i.e. r vs. K, Boyce, 1984). 327 This may indicate that evolutionary responses to invasive predators in natural systems can be              Lake Kegonsa (Table 3). Post-and pre-invasion populations consist of clones originating from 577 after and before Bythotrephes longimanus invasion, respectively.