Understanding how species can thrive in a range of environments is a central challenge for evolutionary ecology. There is strong evidence for local adaptation along large-scale ecological clines in insects. However, potential adaptation among neighbouring populations differing in their environment has been studied much less. We used RAD-sequencing to quantify genetic divergence and clustering of ten populations of the field cricket Gryllus campestris in the Cantabrian Mountains of northern Spain, and an outgroup on the coastal plain. Our populations were chosen to represent replicate high and low altitude habitats. We identified genetic clusters that include both high and low altitude populations indicating that the two habitat types do not hold ancestrally distinct lineages. Using common-garden rearing experiments to remove environmental effects, we found evidence for differences between high and low altitude populations in physiological and life-history traits. As predicted by the local adaptation hypothesis, crickets with parents from cooler (high altitude) populations recovered from periods of extreme cooling more rapidly than those with parents from warmer (low altitude) populations.  Growth rates also differed between offspring from high and low altitude populations. However, contrary to our prediction that crickets from high altitudes would grow faster, the most striking difference was that at high temperatures, growth was fastest in individuals from low altitudes. Our findings reveal that populations a few tens of kilometres apart have independently evolved adaptations to their environment. This suggests that local adaptation in a range of traits may be commonplace even in mobile invertebrates at scales of a small fraction of species’ distributions.

Sampling regime

G. campestris were collected from locations in Northern Spain (Table 1, Fig. 1).   These crickets have a single annual generation, and in this region, which is towards the southerly limit of their range, they live from sea level up to around 1500 m.  Winter snowfalls can occur at low altitudes and high altitude sites are covered in snow for several months in winter, during which time the crickets have an obligatory diapause as late-instar nymphs.  Summer temperatures can reach the high 30s °C at low altitudes.  The mean height (± SD) of low altitude populations is 108±54m and of high altitude populations is 1160±108m.  We do not have precise climatological data for collection sites, but the lower air density is expected to reduce the air temperature by ~ 7°C at high vs. low altitude sites. Mean daily temperatures in 2018 at a site at 127m (Asturias airport (N 43 33.74, W6 02.00)) and one at 1480m (Valgrande ski station (N 42 58.89, 5 46.54)) were 14.2 and 7.2 ºC respectively (datosclima.es, 2020) which suggests a temperature difference between our high and low altitude sites in the region of 5-6°C. The first adults appear in late April or early May at low altitudes and a few weeks later at high altitudes.  The longest lived adults live until late June.

Fig. 1.  Locations of cricket populations near the north coast of Spain.  Populations prefixed with an ‘L’ are at low altitudes, populations prefixed with an ‘H’ are at high altitudes (see table 1). Populations sampled for the common-garden rearing studies in this data set are shown as ● or ◐. Grey shaded areas are land above 1600m, lines south of the coast are major rivers.

Table 1. Sampling locations (see Fig. 1).

Data collection methods for chill coma recovery data

In late April and early May 2018 we collected ten adult or last-instar crickets of each sex from each of populations L1-L5 and H6-H10 using burrow traps (Meadows, 2013).  We took them back to our facility at site L15, and then via Air (in hand luggage) to our lab at the University of Exeter’s Penryn campus, UK. Crickets were maintained in individual 7x7cm boxes provided with ad-lib food (standard rodent diet) and continuous access to water through a 5ml glass vial with a cotton wool stopper in each box.  Boxes were kept in a single temperature-controlled room at 25°C with a 7am-10pm light cycle reflecting the contemporaneous day length in Asturias.  Once adult, females were each mated to a male from their own population, by placing a male in the female’s box and observing them until mating occurred. The position of the boxes was rearranged at least weekly to ensure that potential temperature gradients within the room could not systematically affect populations.  Once females from all populations had been adult for at least 3 days, and at least 5 females per population had mated successfully, all females were provided with wet sand for egg laying.  It is possible that a small proportion of females had mated in the field prior to their laboratory mating.  Because females store sperm this means that some families that we subsequently assume to be full-siblings will include some half-siblings, however, this is conservative in relation to all our analyses.  Females were left for a period of 48 hours before eggs were collected and transferred to a petri dish containing cotton-wool saturated with water and maintained at 25°C in an incubator (1 petri dish per female). Eggs were spread out across the petri dish to reduce the potential for mould to spread between them. As nymphs hatched they were transferred to boxes of the same design as those used for adults at a density of 15 individuals per box. Individuals could not be individually marked, so although the range of hatching dates within a box were known, individual ages were not retained. 

Chill coma assays were run in simultaneous batches of 10 nymphs, with only a single individual taken from any one family and a random mixture of populations in each assay.  When selecting an individual from a box of 15 nymphs, we chose an individual from the middle of the size distribution in the box. Before being tested, nymphs were weighed on a Mettler Toledo ME54E analytical balance (www.mt.com) with a repeatability of 0.1mg.  Each nymph was placed in a 10ml individual vial which was then attached to piece of wood by its lid and the vials then submerged in an ice bath with a mean temperature of 0.4°C for a 1 hour cooling period.

We used a thermometer to monitor the water temperature, adding ice to the water bath if required to maintain the temperature between 0 and 1°C. After 1 hour the vials were removed from the ice bath and crickets tipped onto their backs into individual petri dishes in a 25 °C CT room with typical indoor laboratory/office lighting levels. If any individual(s) landed the right way up, a small paintbrush was used to flip them over (if any nymphs needed to be flipped over, all nymphs in that trial were also brushed).  The time taken for each individual to return to an upright position (CCR time) was measured by two observers watching 5 crickets each and recording the time to the nearest second using a voice recorder. Observers were blind to the altitude of the populations that were tested.  The process was repeated over 5 days measuring 448 nymphs between 4 and 14 days old (age group 1). To increase our sample size, the procedure was repeated with 300 nymphs of 21 to 35 days old (age group 2) and with 205 nymphs of 109 to 120 days old (age group 3) using unique nymphs for all assays.  Some small nymphs (in ages groups 1 and 2) failed to recover from the chill procedure, these dead individuals were excluded from analysis of recovery time, but they are included in the attached dataset where their recovery time is listed as NA.  The third assay was  conducted in our lab at the University of Aarhus (because of the availability of personnel), using identical procedures to those in the UK apart from slightly greater variation in the temperature of the room where the assays were carried out, which ranged between 25 - 28 °C in Aarhus.

Data collection methods for growth rate data

In late April and early May 2019 In spring 2019 we collected 440 wild field crickets Gryllus campestris (198 males and 242 females) from each of populations L1-L5 and H6-H10 using burrow traps (Meadows, 2013) and took them back to the lab at the University of Exeter’s Penryn campus, UK. Crickets were maintained in individual 7x7cm boxes provided with ad-lib food (standard rodent diet) and continuous access to water through a 5ml glass vial with a cotton wool stopper in each box.  Boxes were kept in a single temperature-controlled room at 25°C with a 7am-10pm light cycle reflecting the contemporaneous day length in Asturias.  Once adult, females were each mated to a male from their own population, by placing a male in the female’s box and observing them until mating occurred.  The position of the boxes was rearranged at least weekly to ensure that potential temperature gradients within the room could not systematically affect populations.  Once females from all populations had been adult for at least 3 days, and at least 5 females per population had mated successfully, all females were provided with wet sand for egg laying.  It is possible that a small proportion of females had mated in the field prior to their laboratory mating.  Because females store sperm this means that some families that we subsequently assume to be full-siblings will include some half-siblings, however, this is conservative in relation to all our analyses.  Females were left for a period of 48 hours before eggs were collected and transferred to a petri dish containing cotton-wool saturated with water and maintained at 25°C in an incubator (1 petri dish per female). Eggs were spread out across the petri dish to reduce the potential for mould to spread between them. As nymphs hatched they were each transferred to an individual cage, allowing us to retain information about parentage which is recorded recorded in the ‘Pedigree’ sheet of the attached dataset.   Nymphs began to hatch on the 22/05/19, and we continued to collect them until we reached our target of 1000 nymphs with a maximum of 10 hatchlings from any individual female.  Each nymph was placed into an individual 7x7cm plastic box with a perforated lid for aeration, and a piece of cardboard egg carton as a substrate.  They were provided with a water vial (as described above) and an excess of a mixture of two ground up foods (‘Pets at Home Rat Nuggets’ (protein 16%, crude fibre 4%, crude fat 5%, crude ash 5%) and ‘Burgess Excel Nature’s Blend Rabbit Nuggets’ (protein 12.6%, crude fibre 19%, crude fat 3.6%, crude ash 6.5%)). Each nymph box was labelled with only a number to ensure the study was conducted blind and was allocated to one of two temperatures 23°C or 28°C.  These temperatures were chosen because our pilot studies showed very low growth rates below 21°C and very high mortality above 30°C.  Also the 5 degree temperature difference is in the region of the 7 degree difference that we expect to find between high and low altitudes due to the difference in air density. Boxes were distributed among 2 incubators at each temperature with L:D 18:6 photoperiod cycles, and 50% humidity. Boxes were moved around within and between the 2 incubators at each temperature every few days. Each individual cricket was weighed within 24 hours of hatching (before it had access to any food), and every 13 days until 39 days later, using a Mettler Toledo XS3DU analytical balance (www.mt.com) with a listed repeatability of 1µg.   If the nymph died it is recorded as NA in the datasheet; only superficially healthy, living crickets were weighed.

Laboratory instrumentation

We recorded temperature in the laboratory using a variety of generic laboratory thermocouple-based thermometers with accuracies of +- 1 degree C.  For chill coma assays, nymphs were weighed on a Mettler Toledo ME54E analytical balance (www.mt.com) with a repeatability of 0.1mg.  For growth rate assays nymphs were weighed on a Mettler Toledo XS3DU analytical balance (www.mt.com) with a listed repeatability of 1µg.

Quality control

To determine repeatability of mass measurements, we weighed 38 newly hatched nymphs twice, yielding a repeatability of 0.96.  Standard statistical approaches for identifying errors in data entry were applied to the data, no anomalous values were identified, no data points were removed from the raw dataset.  NA values in the dataset are nymphs that died during the experiment.  For the chill coma recovery data, standard statistical approaches for identifying errors in data entry were applied to the data, a single cricket was recorded as having a mass of 10 grams which is 20x  the mass of any other cricket and entirely impossible.  This was assumed to be a data entry mistake and the individual was removed from the dataset.