Cascading impacts of earthquakes and extreme heatwaves have destroyed populations of an iconic marine foundation species

Ecologists traditionally study how contemporary local processes, such as biological interactions and physical stressors, affect the distribution and abundance of organisms. By comparison, biogeographers study the distribution of the same organisms, but focus on historic, larger‐scale processes that can cause mass mortalities, such as earthquakes. Here we document cascading effects of rare biogeographical (seismic) and more common ecological (temperature‐related) processes on the distribution and abundances of coastal foundation species.


| INTRODUC TI ON
Ecologists traditionally study how ubiquitous processes such as competition, predation, disturbance and abiotic stress affect the distributions and abundances of populations, species and communities (Begon et al., 1986;Smith & Smith, 2015). These processes typically take place over short temporal (days-decades) and small spatial (cm-km) scales. Similarly, larger-scale events, such as hurricanes, fires, droughts and heatwaves, which are often associated with climate change, can have important effects on ecosystem structure (Buma, 2015;Lugo, 2008;Lundholm, 2009;Sippel et al., 2018).
For example, atmospheric heatwaves have altered population demographics and species interactions and caused poleward range movements and localized extinctions of foundation species, which provide critical biogenic habitat for associated species and ecosystem functioning (Chen et al., 2011;Dillon et al., 2010;Ellison, 2019;Thomas et al., 2004). Like ecologists, palaeontologists and biogeographers also study processes that affect distributions and abundances of organisms but focus on rarer events and longer temporal (millennia to millions of years) and larger spatial (km to global) scales, such as stochastic long-range dispersal incidents or cataclysmic tectonic events that can cause extinctions and dispersal barriers (Black & Black, 1988;Brown, 2009;Cox et al., 2016;Crisci, 2001).
However, these traditional disciplinary boundaries are humanmade constructs, and it is increasingly recognized that studies across spatiotemporal scales are required for better understanding of present-day distributional patterns (Crisci et al., 2006;Wiens & Donoghue, 2004).
In the marine environment, heatwaves have intensified in recent decades, and models project that they will continue to become stronger and more frequent (Frölicher et al., 2018;Hobday et al., 2018;Oliver et al., 2019;Sen Gupta et al., 2020). Over the last few decades, marine heatwaves have caused coral bleaching, localized extinctions, poleward range-shifts of many species of fish and seaweed and altered species interactions (Cheung & Frölicher, 2020;Wernberg et al., 2013).
At longer time scales, many coastal ecosystems have also experienced extreme disturbances associated with seismic activity such as earthquakes and volcanic eruptions (Castilla & Oliva, 1990;Castilla et al., 2010). However, traditional marine ecology (e.g. textbooks like Castro & Huber, 2019;Kaiser et al., 2011) do not typically discuss large-scale disturbances associated with seismic activity in a context of contemporary species distributions and biodiversity.
Nevertheless, rare mega-disturbances do affect local biodiversity with ramifications for many years to come. For example, historic and contemporary earthquakes have affected present-day species distributions (Haven, 1964;Noda et al., 2016;Rodil et al., 2016). Such examples show that the extreme events traditionally studied by biogeographers and palaeontologists can be relevant for ecologists and the study of contemporary ecosystems.
On November 14, 2016, a 7.8 M w (on the moment magnitude scale) earthquake struck the northeast South Island of New Zealand, with the epicentre located 60 km south-west of the small coastal town of Kaikōura Hamling et al., 2017;Kaiser et al., 2017). With a shallow hypocentre of 15 km and many complex inshore and offshore faults, slips, vertical displacements, coastal uplifts of up to 6.5 m, and >2,000 aftershocks in only 3 days, four of which had magnitude >6 M w , these earthquakes directly affected c. 130 km coastline Hamling et al., 2017;Xu et al., 2018). Over the next few months, we observed extensive loss of habitat-forming seaweeds and slow-moving benthic invertebrates, with associated losses of primary productivity, biogenic habitat and altered food webs Schiel et al., 2019;Thomsen et al., 2020). A year later, over the austral summer of 2017/18, New Zealand experienced the strongest marine heatwave on record (Salinger et al., 2019(Salinger et al., , 2020. This large-scale extreme marine event, coincided with high air temperatures, low tides and calm sea conditions (Salinger et al., 2020;, and had widespread effects such as high glacial melting, losses of habitat-forming seaweeds, and movement of fish into warm waters (Salinger et al., 2020;Schiel et al., 2019;Tait et al., 2021;. Following Hobday et al. (2018), this large-scale temperature anomaly was named the "Tasman Sea 2017/18 marine heatwave" Salinger et al., 2019). However, analyses of local sea temperature data show that this heatwave was comprised of multiple consecutive events (see Result section), and the term is hereafter pluralized and referred to as "heatwaves." Some of the major coastal species that appeared to be immediately affected by the earthquakes were the southern bull kelps, Durvillaea spp. (see Figure 1, hereafter just bull kelps). Bull kelps are the world's largest fucoid algae, with individuals of some species reaching up to 10 m long, weighing up to 70 kg and living up to 10 years (Hay, 1979, Nelson 2013, Hay, 2020. Bull kelps are dominant conspicuous habitat-forming seaweeds on the intertidalsubtidal fringe of rugged wave-exposed reefs in New Zealand and many other places in the southern hemisphere where they control local diversity and ecological functioning (Schiel, 2019;Taylor & Schiel, 2005;. Southern bull kelps are also iconic, culturally important seaweeds in New Zealand and other parts of the world, such as Chile. For example, there is a long traditional use of Durvillaea spp. as storage bags, known as "pōhā" in New Zealand, and as a source of food in Chile for at least 14,000 years (Dillehay et al., 2008). The iconic nature of bull kelp is reflected in its use in cultural and spiritual symbolism (Pérez-Lloréns et al., 2020) such as being emblematic of the knowledge and status of the people K E Y W O R D S alternative foundation species, cataclysmic disturbances, habitat-formers, regime shift, seismic uplift, turf algae of Kaikōura (Anon., 2007). We examined changes to the distribution and abundance of bull kelp around Kaikōura, following the 2016 earthquakes and 2017/18 heatwaves and whether they returned or were replaced by alternative foundation species (large canopyforming seaweed, such as Lessonia variegata and Carpophyllum maschalocarpum) that were also common in the region (Schiel, 2006;. These results are discussed in the context of other large-scale impacts over long time periods.

| Study system and marine heatwaves in the Kaikōura region
The bull kelp species in this study were Durvillaea antarctica, which occurs at the lowest tidal zone, and D. willana, which occurs slightly deeper in the subtidal zone and is only partially exposed at the lowest tides. A third bull kelp species, D. poha, can be morphologically similar to D. antarctica, but inhabits slightly less wave-exposed habitats and is absent (or very rare) along the Kaikōura coastline (Fraser et al., 2012;Peters et al., 2020;Vaux et al., 2021;Velásquez et al., 2020). Additionally, there are two recognized clades of D. antarctica in New Zealand that are distributed to the north and south of Banks Peninsula, and therefore, most of the D. antarctica populations studied here belonged to the D. antarctica "NZ north" clade, whereas the Moeraki samples used as controls in survey 1 were of the D. antarctica "NZ south" clade . Bull kelps have exceptionally strong and large holdfasts and their heavy fronds affect the subcanopy environment and biodiversity through shading and whiplash (Schiel, 2019;. The physical attributes of the Tasman Sea 17/18 marine heatwaves, and covarying environmental factors, such as low wind speed, fewer waves and high land temperatures, as well as co-occurring spring tides, have been analysed and discussed in great detail Salinger et al., 2019Salinger et al., , 2020 The distance from the reef sites to the grid centre was <25 km. For each of the two sites, we identified all MHWs and calculated their maximum intensity (temperature above the 90% threshold in °C), durations (in days) and cumulative intensity (duration × intensity).
For brevity, we only show maximum intensity (Figure 2b), whereas duration and accumulated intensity are reported in the Appendix S1.
The MHW analyses were done using the R package heatwaver version 0.4.5 (Schlegel & Smit, 2018;Smit et al., 2018), which defines a MHW as a period of 5 or more consecutive days where the sea surface temperature is greater than the 90th percentile calculated from a 30 year climatology (period between 1/1-1983 and 31/12-2012).
Analyses were based on NOAA high-resolution blended analysis of daily sea surface temperature data in ¼ degree grids derived from satellites and in situ data (oisst v2.1, accessed from https://coast watch.pfeg.noaa.gov/erdda p/index.html) (Huang et al., 2021). More details, including codes and outputs, are available at: https://rpubs. com/FranT oto/Thoms en2021_MHW and https://github.com/FranT oto/Thoms en_etal_2021_MHW_EQ. Maximum intensity of marine heatwaves recorded offshore of the Kaikōura Peninsula and Oaro between 1982 and 2020 (see Appendix S1 for similar data for other heatwave metrics). (c) Sample sites along the Kaikōura coastline. Oaro is the non-uplifted control reef, and the blue markers show locations of 15 uplifted reefs (0.5-2 m uplift) numbered with distance from Oaro. All reefs were sampled in Survey 1, reefs 1, 2, 4, 6, 7 and 8 were sampled in Survey 2, and reefs 1, 2, 4, 6, 9, and 15 were sampled in Survey 3 The control reef at Oaro was selected due to its proximity to uplifted reefs and because it had similar high cover of bull kelp beds prior to the earthquake (Figure 1a). Oaro and the 15 uplifted reefs (0.5-2 m), were surveyed in February-March 2017 during low spring tides, when access was logistically possible. For this study, the zone that had been inhabited by bull kelp prior to the uplift was divided into three elevation bands. The high band was visibly "white" because all the calcified understory red algae had died and were bleached by the sun (Figure 1b-f). There were also many obvious holdfast scars of dead bull kelp in this zone (Figure 1d,e). The mid band was a bright "green" because it was dominated by the ephemeral green algae Ulva spp. (Figure 1b,g) Schiel et al., 2019).
In this band, most Durvillaea holdfasts were still present with their stipes or stipe-remnants ( Figure 1f,g), but their fronds had decomposed (i.e. "ghost-holdfasts" as described in . Previous analyses have shown that small-scale, reef-specific degree of uplift did not correlate with any of the five measured responses (Mondardini, 2018) and this aspect was therefore not explored.
None of the responses could be transformed to variance homogeneity (Levine's tests, p < .001), and sampling was unbalanced between elevation bands and uplifted and control reefs. Standard or permutational based ANOVA could therefore not be used (Anderson et al., 2017). Instead, we used pairwise Mann-Whitney tests on relevant data subsets to test for differences between the control site at Oaro versus uplifted reefs (a total of 10 tests for the 5 responses; excluding the high band, which was not present at Oaro) and between the three elevation bands at the uplifted reefs only (a total of 15 tests for the 5 responses). We note that 1 in 20 tests may be inflated based on an a priori significant level of 0.05 when multiple analyses are done (Anderson, 2005). All bull kelp and algae data are reported as means and standard errors (SE).
To relate abundances of bull kelp holdfasts (with or without blades as measured above) to pre-earthquake healthy live bull kelp canopies, percent cover of bull kelp holdfasts was estimated in 40 50 × 50 cm quadrats haphazardly positioned within dense bull kelp beds (i.e. with 100% canopy cover) at Moeraki, a southern site not affected by earthquakes and less affected by the Tasman Sea 17/18 heatwave . The 40 quadrats were collected over a period of 2 years with ten quadrats collected in Autumn and Spring in 2019 and 2020.

| Survey 2: Cascading loss of bull kelp after the earthquakes and heatwaves
Six of the uplifted reefs ( Figure

| Survey 3: Recovery or replacement of bull kelp with alternative foundation species after 4 years
Bull kelps are competitively superior foundation species that control biodiversity (Schiel, 2019). However, there are a range of species on the Kaikōura coastline that are also foundation species and, although they are typically subordinate to bull kelps, it is possible that they benefited from large-scale losses of bull kelp to become more dominant. We evaluated whether bull kelp had recovered or been replaced and other brown algae. Survey 3 data were only evaluated graphically due to the absence of "before" data and because statistical tests among shore bands were unnecessary due the almost total absence of alternative foundation species in the high and mid bands.

| Earthquakes and heatwaves in a global context
To consider our results in a global context, rare and potentially cata- from each seismic event to the nearest coastline was calculated.
We also quantified the extent of coastline affected by of each of the 62 most extreme marine heatwaves (covering global satellite data from 1982 to 2017). First, shape-polygons were extracted from Figure 5 in Sen Gupta et al. (2020). These were overlaid onto a map of the world's coastline, converted to raster-format and the distance of coastline affected by each extreme marine heatwave was calculated.
The same methodology described in Sen Gupta et al. (2020)

| Marine heatwaves in the Kaikōura region
The

| Survey 1: Loss of bull kelp after the earthquakes
There were significant differences in the low band between the uplifted reefs and the Oaro control reef, for all response variables (p < .001), except "stipe with blades" (p = .50, Figure 3). However, in the mid band only "stipes without blades" (p = .001) and "holdfast percentage cover" (p = .027) were significantly different between

| Survey 2: Cascading loss of bull kelp after the earthquakes and heatwaves
Loss of bull kelp after the earthquake was significantly greater at the uplifted reefs compared to at Oaro (

| Survey 3: Replacement of bull kelp with alternative foundation species
There were no bull kelps in the high band, a few scattered individuals in the mid band, and low cover in the low band ( Figure 5, 1.5% D. antarctica, 7% D. willana). By contrast, small foliose, filamentous, and turf seaweeds, typically between 3 and 30 cm long, dominated the low and mid-elevation bands. Encrusting algae were found in all bands but were less common in the low band (4% cover) compared to healthy bull kelp beds (typically c. 60% cover, see supporting data).
Six alternative foundation species were found in the mid (with low cover values) and low elevation bands (with low to mid-cover values) including Carpophyllum maschalocarpum (most common), followed

| Loss of bull kelp
The earthquakes in November 2016 uplifted ca. 50 km of waveexposed rocky substrate between 0.5 and 6 m (31 km of boulder reefs and 16 km of consolidated reef) . This uplift caused a 75% canopy loss of bull kelp in the southern Kaikōura region. Furthermore, we found an additional 35% canopy loss of the surviving bull kelp populations following consecutive heatwaves over the summer of 2017/18. Dramatic uplift-associated loss has also been observed on waveprotected reef-platforms for smaller intertidal habitat-forming fucoids, such as Hormosira banksii and Cystophora spp., following the same earthquakes Thomsen et al., 2020). It is likely that the 75% loss of bull kelp encapsulates a disproportional amount of the intertidal D. antarctica compared to the shallow subtidal D. willana (Figures 1h and 5)  inhabits depths of 0-5 m, and therefore, many individuals have remained within the vertical range of the species after the 0.5-2 m uplift at the study sites (Hay, 1979;Vaux et al., 2021). Greater rates of heatwave-associated loss of bull kelp were also proportionally less on the uplifted reefs compared to Oaro (Figure 4), perhaps because the uplift had already selected hardier individuals (Bennett et al., 2015;Gurgel et al., 2020;King et al., 2018) or possibly because the bull kelp beds contained D. poha, a species of bull kelp with a typically southern distribution and likely less tolerant to temperature stress Vaux et al., 2021;Velásquez et al., 2020). Together, the earthquakes and heatwaves have massively reduced the abundance of bull kelp along the Kaikōura coastline, ranging from heatwave only induced losses at non-uplifted sites (>60% at Oaro) to beyond the extent quantified in this study (north of Kaikoura peninsula to Cape Campbell) where reefs were uplifted up to 6 m and almost all bull kelp were destroyed Schiel et al., 2019;Thomsen et al., 2020).
To date, most research on impacts from marine heatwaves has focused on the effects of temperature on subtidal algal forests in isolation Straub et al., 2019). However, grazing, wave exposure, turbidity or nutrient levels can modify temperatureinduced losses of kelp (Butler et al., 2020;Ling et al., 2009;Tait et al, 2021;Zimmerman & Robertson, 1985). Multiple co-occurring and cascading ecological stressors and disturbances often have complex interactions and it is therefore important to quantify impacts from marine heatwaves in concert with other stressors (Crain et al., 2008;Harvey et al., 2013;Hawkins et al., 2008). Here, bull kelp losses from heatwaves were smaller on reefs that had already experienced losses due to seismic uplift, but these results could have been modified by other covarying stressors such as high summer air temperatures, high irradiance, lower-than-usual tides and low wave energy events (Salinger et al., 2019(Salinger et al., , 2020. In other words, impacts from marine heatwaves may often be modified by a complex set of covarying factors, and therefore, they should be studied as a multi-factorial stressor.

| Wider ecological consequences
While it is challenging to determine the exact areal extent of bull kelp loss following the earthquakes, preliminary estimates suggest that around 125,000 m 2 bull kelp forest could have been lost to the uplift (if 25% of the rocky coastline was dominated by bull kelp and that the combined mid and high bull kelp band on average was ca. 10 m wide).
The wider cascading impact from the uplift and heatwaves have resulted in the loss of millions of bull kelp individuals and likely an ongoing reduction in bull kelp cover on the Kaikōura coast (Figure 7).
Surviving smaller and more patchy populations can have lower genetic diversity and may be less resilient to recover from future disturbances (Buma, 2015;Elmqvist et al., 2003;Frankham, 2005).
Furthermore, new ecological states can arise when severe or cumulative disturbances serve as tipping points from one state to another and alter patterns of recruitment, habitat dominance and networks of interactions within an ecosystem (Benedetti-Cecchi et al., 2015;Dai et al., 2012;Hawkins et al., 2015;Moore, 2018). It is likely that the positive feedbacks (propagule pressure, habitat maintenance) that maintained bull kelp forests have been lost and it is possible that a new stable state dominated by small turf and foliose algae is in development (Figure 7), as has been seen in many other parts of the world (Filbee-Dexter et al., 2016;O'Brien & Scheibling, 2018). Such turf assemblages are typically limited by large canopy-forming algae, but once established can prohibit colonization by canopy-formers through habitat modification, competition for primary space and the resulting reductions in propagule pressure (Jenkins et al., 2004;Kennelly, 1987;Petraitis & Dudgeon, 2004;Smale, 2020). If such an alternative state persists, it will likely result in long-term reductions of bull kelp-associated species and ecosystem services such as carbon-storage and the dampening of wave action (Filbee-Dexter & Wernberg, 2018;Smale, 2020).

| Caveats and limitations of the study
The inherent unpredictability of cataclysmic events such as earthquakes often necessitates ad hoc research designs such as the one presented here. In this instance, we had very few data for the most impacted zones prior to the earthquake, which would have allowed for direct before/after contrasts. Instead, we developed a method based on robust ecological criteria (presence of dead, decaying and living bull kelp holdfasts) to estimate the extent of bull kelp cover prior to the earthquakes and heatwaves at our study sites. The tenacity of the bull kelp holdfasts, contrasts to an unimpacted control site, ancillary data from a southern reference site, and our extensive experience working on the Kaikōura coastline Schiel et al., 2016;Thomsen et al., 2020) contributed to this being a robust, if not ideal method. For example, it was impossible to determine whether decayed holdfasts were D. willana or D. antarctica, preventing us from assessing the pre-and post-earthquake relative abundance of these species at our study sites. Another limitation of this study was the lack of multiple control sites to incorporate sitesite variability into our contrasts between uplifted and non-impacted reefs. However, our control reef at Oaro was the only unimpacted reef along ~100 km of coastline, and the only accessible bull kelp reef for hundreds of kilometres and is typical of bull kelp reefs elsewhere (Schiel, 2019;Schiel et al., 2018;.

| The Kaikōura earthquakes and heatwaves in a global context
Earthquakes and extreme heatwaves are often considered unique and rare disturbances that have little relevance for traditional and contemporary ecology (e.g. textbooks such as Begon et al., 1986;Castro & Huber, 2019;Kaiser et al., 2011;Nybakken, 1993;Smith & Smith, 2015). However, on a global scale almost 15,000 high-impact F I G U R E 7 Conceptual diagram showing impacts from the 2016 earthquakes and 2017/18 heatwaves on many uplifted reefs along the Kaikoura coastline. (a, b) Dense forests of 2-8 m large bull kelps are maintained by positive feedbacks characterized by high propagule production, refugium from consumers, chemical cues and frond abrasion to facilitate recruitment and growth of new bull kelp, abalone, crayfish, fish and holdfast fauna. (c) The earthquakes and heatwaves destroyed many of these forests. (d) Dense bull kelp forests have been partly replaced by 0.05-0.3 m foliose and turfing algae, with interspersed patches of surviving bull kelp and alternative foundation species such as Carpophyllum spp. and Cystophora spp. The animal communities, feedback mechanisms, stability of the new system and mechanisms that may tip it back to bull kelp forests remain unknown. The red arrow represents a tipping point and state-change whereas the blue arrows show possible feedback mechanisms that maintain a state earthquakes and volcanic eruptions have been recorded over the last 4 millennia, of which 75% were within 100 km of the coastline, providing circumstantial evidence that seismic activity may have common legacy effects. While the individual traits (e.g. uplift, subsidence or horizontal displacement) of these 15,000 earthquakes have not been studied, it is likely that their impacts varied depending on their traits and magnitudes, and the physical and biological characteristics of the affected shoreline. For example, earthquakes can directly affect organisms by altering their elevation on a shore, through indirect effects such as increased sedimentation, but also through the destruction and creation of new habitat, as occurs when boulders are deposited in the marine environment (Bodin & Klinger, 1986;Castilla et al., 2010;Schiel et al., 2019;Thomsen et al., 2020;Vaux et al., 2021). The Kaikōura earthquake provided evidence to support recent research that has implicated historic seismic events in the contemporary distribution of bull kelps in New Zealand Hay, 2020;Parvizi et al., 2020;Vaux et al., 2021). It is likely that strong seismic events have been important structural forces that underly present-day coastal ecology in many parts of the world.
Superimposed on dramatic geological events are an increasing number of unusually hot oceanic conditions that are caused by heatwaves Sen Gupta et al., 2020). Temperature affects all aspects of biology, from biochemical rates at subcellular levels, reproduction rates, control over species ranges and ultimately the distribution of world's major biomes (Bartsch et al., 2012;Lüning, 1990;Spalding et al., 2007). It is therefore not surprising that strong heatwaves have altered the ecology of impacted regions (Rogers- Bennett & Catton, 2019;Smale et al., 2019;Straub et al., 2019). Globally, we estimated that the most extreme of these events have caused elevated temperatures across ca. 182,300 km of coastline (ca. 15% of global coastline see Results) since 1982. A few of these events have been studied in detail, demonstrating significant, and often detrimental, impacts on local marine biota (Jones et al., 2018;Montie et al., 2020;Rogers-Bennett & Catton, 2019;Smale et al., 2019;Wernberg et al., 2016), but most events have simply not been studied.
Many types of large-scale extreme disturbances can create similar ecological legacy effects and are likely important drivers of contemporary species-distribution patterns. For example, extreme 1000-year floods, fires and hurricanes can devastate coastal communities through extreme run-off, water turbidity, enhanced sedimentation, lowered salinity, wave action, altered biogeochemistry and through ash-deposits (Dunbar & McCullough, 2012;Ely et al., 1993;Flannigan et al., 2006;Kunkel et al., 2013). Furthermore, some of these events are more important on coastlines that have low seismic activity (i.e. coastlines that appear less affected in Figure 6).
The possibility that many of these events will become stronger and more frequent in the future, highlights their importance in both ecology and biogeography and calls for greater inter-disciplinary cross-scale approaches (Alfieri et al., 2017;Ely et al., 1993;Flannigan et al., 2006;Walsh & Pittock, 1998).

| CON CLUS ION
This study demonstrated that cumulative effects from seismic uplift and a subsequent heatwave caused great mortality of a marine foundation species. The uplift first caused 75% canopy loss of bull kelp, and then, the heatwaves killed an additional 35% of those remaining, resulting in cascading losses of likely millions of individuals on the Kaikōura coast. Four years after the uplift, bull kelp had not recovered and the low zone was instead inhabited by a mixture of patchy bull kelp beds (dominated by D. willana), a few other perennial habitat-forming species such as Carpophyllum maschalocarpum, and small ephemeral turf and foliose algae. This represented a different ecological system maintained by new feedback loops that likely slow down or hinder recovery of bull kelp beds and lower their resilience to future stressors. Cataclysmic events may be relatively common on global historical scales because more than 12,000 major events have been recorded within 100 km of the coastline over the last four millennia, and because relatively recent extreme marine heatwaves have occurred along more than 180,000 km coastline or about 15% of global coastline around the world. Thus, the ecological legacy effects of large-scale disturbances, such as earthquakes, heatwaves and other extreme events, may be relatively common. It is worthwhile viewing local-scale short-term ecological research in this wider historical context. Models that incorporate appropriate legacy effects will likely yield better interpretation of contemporary local-scale processes.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflicts of interest.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/ddi.13407.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data presented in this study are openly available in DRYAD at https://doi.org/10.5061/dryad.v6wwp zgwq.