Spatial selection and high relatedness

at range fronts collectively increase

the rate of range expansion

Abstract

Global change is leading to significant changes in species distributions. Understanding the processes at play during range shifts is therefore of paramount importance. Studies increasingly exemplify the necessity to take into account evolutionary change when examining such shifts. Spatial selection and local adaptation have indeed been shown to both shape phenotypic differentiation along expansion gradients. While acknowledging this importance of evolution, we here argue that plasticity (in particular the condition-dependency of dispersal behaviour) may be an overlooked (co-)driver of range expansions. Using an approach of experimental evolution within microcosms, we evaluated the relative importance of spatial selection and kin competition in pushing range advance. As a model species, we used the spider mite Tetranychus urticae. In a first experiment, a control was compared with a treatment where mites were regularly reshuffled to nullify any spatial sorting of phenotypes and any spatial structure in relatedness. In a second, a control was compared with a treatment using inbred mites. Where both reshuffling and the use of inbred mites significantly slowed down range advance compared to the controls (suggesting spatial selection), the reduction through reshuffling was more than double. This indicates that the destruction of the spatial structure in relatedness (i.e. increased kin competition near the front) had a bigger effect than the impediment of evolutionary change. Our results thus provide the first empirical evidence of kin competition as an overlooked but significant proximate driver of range expansion. While the loss of genetic variation during range expansion is typically considered to be an eventual constraining factor because it limits evolutionary potential, we here show that increased relatedness may in contrast plastically increase the rate of range advance.

Introduction

In the context of contemporary global change, understanding the processes shaping species’ ranges has become increasingly important. Climate change and species introductions, for example, lead to changes in species distributions (Parmesan 2006) and thus altered species interactions (e.g. Moorcroft et al. 2006; Pateman et al. 2012). This will ultimately affect community composition (Ko et al. 2014; Barbet- Massin & Jetz 2015) and hence ecosystem functioning. Apart from the ecological costs, changed species distributions may moreover imply severe economic costs (e.g. invasive pest species Pimentel et al. 2005; Aukema et al. 2011).

Accumulating empirical and theoretical studies over a wide range of organisms demonstrate increased dispersal rates at range expansion fronts (e.g. Travis & Dytham 2002; Hughes et al. 2007; Perkins et al. 2013; Hargreaves & Eckert 2014; Van Petegem et al. 2016). Where increased dispersal at range margins can largely be understood within an ecological framework (where biotic interactions and the spatial and temporal features of the abiotic environment all influence dispersal behaviour, see Kubisch et al. 2014), a pure ecological emphasis may fall short within the context of actively expanding ranges as it ignores the potential for evolutionary change. Increasingly accelerated rates of range advance over time (e.g. Travis & Dytham 2002; Lindstrom et al. 2013), for example, may ask for explanations that involve evolution. Present- day research therefore often has a very strong focus on this importance of evolution in shaping traits during range expansion. Indeed, today, increased rates of range advance are mainly explained through the evolutionary process of spatial selection (Phillips et al. 2010; Shine et al. 2011). Central to this process are the spatial assortment of dispersal phenotypes through space and a shift from K- towards r-selection: Firstly, as they will always be the first ones to arrive, the most dispersive

phenotypes typically accumulate at the range front and mate amongst each other to produce on average even more dispersive offspring (i.e. the Olympic village effect, see Phillips et al. 2010). Secondly, in the absence of Allee effects, individuals near the leading edge may benefit from the local low-density environment and produce more offspring, resulting in a shift from K-selection in established populations (core) towards r-selection near the front (Phillips 2009) (our use of the r-K selection spectrum thus refers to the idea that density-regulated (core) and density-independent (front) populations typically differ with regard to optimal life-history strategies –cfr. Phillips et al. 2010). Together, this increase in dispersal and reproduction may push the range forwards.

The potential role of plasticity as a (co-)factor pushing range advance is therefore often neglected (but see Pettit et al. 2016). Dispersal behaviour, however, is known to be an informed, highly condition- dependent behaviour that relies on various proximate cues, including local population density and relatedness (Clobert et al. 2009). High densities, for example, lead to more frequent dispersal events over greater distances (Poethke & Hovestadt 2002; Poethke et al. 2011; Bitume et al. 2013; Dahirel et al. 2016). Interestingly, kin competition may lead to similar condition-dependent changes in dispersal behaviour (Bitume et al. 2013; Nitsch et al. 2016) and is predicted to increase towards the leading edge because of serial founder effects, whereby small founder populations rapidly build up high levels of relatedness and thus suffer from increased competition among kin. A high level of relatedness could thus be a potential proximate driver of range advance (in the theoretical study of Kubisch et al. 2013, kin competition was already shown to be an ultimate cause of range advance). To date, however, any empirical evidence on the importance of condition-dependency (density-dependence and kin competition) in explaining range expansions is lacking.

The current study was set up to disentangle between spatial selection and the condition-dependency of dispersal as drivers of

range advance. It thus adds to three recent studies where only the role of spatial selection was evaluated (Ochocki & Miller 2016; Wagner et al. 2016; Williams et al. 2016). Using the two-spotted spider mite Tetranychus urticae Koch (Acari, Tetranychidae) as a model species, we set up two microcosm experiments, where mites were allowed to expand their range within experimental metapopulations. In a first experiment, we compared unmanipulated metapopulations with metapopulations where we nullified potential effects of kin competition (cfr. (Kubisch et al. 2013) and evolution (cfr. Livingston et al. 2012; Ochocki & Miller 2016; Williams et al. 2016) through the random reshuffling of phenotypes. This allowed evaluating the combined role of spatial selection and condition-dependency (relatedness) in pushing range advance. In a second experiment, we compared unmanipulated metapopulations with metapopulations where we impeded evolutionary change by using mite strains depleted of genetic variation (cfr. Turcotte et al. 2011). This allowed evaluating the combined role of spatial selection and condition-dependency versus that of condition-dependency only. Finally, by comparing between our two experiments, we could indirectly evaluate the role of relatedness (difference in kin competition between treatments of first, but not second experiment).

More specifically, we thus (i) surveyed spread dynamics to see if/ which treatments differed in their rate of range advance and (ii) evaluated the role of spatial selection (evolved dispersal and/or r) versus condition- dependency (relatedness) in shaping these dynamics. Furthermore, we (iii) examined whether potential trends in r could be translated to changes in eight individual life-history traits that constitute r.

Materials and methods

Study species

The two-spotted spider mite, Tetranychus urticae Koch (Acari, Tetranychidae) is a herbivorous species with a worldwide distribution. It is considered a major agricultural pest, especially in greenhouses and orchards (Kennedy & Storer 2000; Hill 2008). Over the last thirty years, the species has expanded its European range from the Mediterranean to (at least) southern Scandinavia (personal observation first author) (see also Carbonnelle et al. 2007). T. urticae has the ability to disperse over long distances by actively making use of aerial currents (i.e. aerial dispersal behaviour, cfr. Hoy et al. 1984). Alternatively, it can engage in ambulatory movements for more directed, short-distance dispersal (i.e. ambulatory dispersal behaviour). Under benign temperatures (typically around 28°C), the mite species can develop from egg to adult in about 9 days (Sabelis 1981). The species is highly fecund (Krainacker & Carey 1989) and reproduces through arrhenotokous parthenogenesis. Unfertilised eggs thus develop into males, whereas fertilised eggs develop into females (Helle 1967b). The sex ratio in a population is usually female-biased (3:1, see Krainacker & Carey 1990).

Microcosm experiment

Experimental metapopulations

An experimental metapopulation consisted of a linear system of populations: bean patches (2 x 2 cm2_{) connected by parafilm bridges}

(8 x 1 cm2 _{–cfr. Bitume et al. 2013), placed on top of moist cotton (figure}

1). A metapopulation was initialised by placing ten freshly mated 1-day- old adult females on the first patch (population) of this system. At this point, the metapopulation comprised only four patches. The initial

population of ten females was subsequently left to settle, grow and colonise the next patch(es) in line through ambulatory dispersal. Three times a week, all patches were checked and one/two new patches were added to the system when mites had reached the one but last/last patch. Mites were therefore not hindered in their dispersal attempts, allowing a continuous expansion of the range. A regular food supply was secured for all populations by renewing all patches in the metapopulation once every week: all one week old patches were shifted aside, replacing the two week old patches that were put there the week before, and in their turn replaced by fresh patches. As the old patches slightly overlapped the new, mites could freely move to these new patches. Mites were left in this experimental metapopulation for approximately nine overlapping generations (corresponding with more or less 75-80 days) during which they could gradually expand their range.

Figure 1: artificial metapopulation (microcosm). The metapopulation was initialised with ten