Climate change is acting on species and modifying communities and ecosystems through changes not only with respect to mean abiotic conditions, but also through increases in the frequency and severity of extreme events. Changes in mean aridity associated with climate change can generate ecotype by environment mismatch (i.e., climatic displacement). At the same time, variability around these shifting means is predicted to increase, resulting in more extreme droughts. We characterized the effects of two axes of climate change–climatic displacement and drought–on the shrub Artemisia californica and its arthropods. We established common gardens of plants sourced along an aridity gradient (3.5-fold variation in MAP) in an arid region of the species distribution, thus generating a gradient of climatic displacement (sustained increase in aridity) as predicted with climate change. We surveyed plants and arthropods over eight years where precipitation varied 6-fold, including both extreme drought and relatively mesic conditions. These two axes of climate change interacted to influence plant performance, such that climatically-displaced populations grew slowly regardless of drought and suffered substantial mortality during drought years. Conversely, local populations grew quickly, increased growth during wet years, and had low mortality regardless of drought. Effects on plant annual arthropod yield were negative and additive, with drought effects exceeding that of climatic displacement by 24%. However, for plant lifetime arthropod yield—incorporating effects on both plant growth and survival—climatic displacement exacerbated the negative effects of drought. Collectively these results demonstrate how climatic displacement (through increasing aridity stress) strengthens the negative effects of drought on plants and, indirectly, on arthropods, suggesting the possibility of climate-mediated trophic collapse. --

From Croy et al (2021):

Artemisia californica (Less. Asteraceae) is a dominant shrub of California’s biodiverse and threatened coastal sage scrub ecosystem (Myers et al. 2000) and supports a species-rich arthropod community (Pratt et al. 2017). The species can live up to 25 years (Sawyer et al. 2009) and relies on wind for pollination and seed-dispersal. This shrub spans a 1,000 km distribution that encompasses a five-fold precipitation gradient from Northern Baja, Mexico (average annual precipitation: 20 cm) to Mendocino County, California (average annual precipitation: 103 cm). Studies have documented genetically-based trait variation across populations of A. californica that is suggestive of locally adapted ecotypes (Pratt and Mooney 2013). These ecotypic differences in turn influence the abundance and community composition of arthropods (Pratt et al. 2017) that are both a key component of biodiversity and support several endemic and endangered vertebrates that drive regional conservation efforts (Bowler 2000). Climate projections for the region include both northward shifts in aridity and an increased frequency and severity of droughts (Diffenbaugh et al. 2015, Wang et al. 2017, Swain et al. 2018; but see Wang et al. (2017) on simultaneous projections of increased deluge), and there is evidence this change is already underway (Pratt and Mooney 2013, MacDonald et al. 2016). This current study is based upon populations of A. californica distributed over 700 km in southern and north-central California (32.8-37.8° latitude; 26.6-91.6 cm precipitation) that together represent 67% of its range and include 80% of the precipitation gradient defining its overall distribution.

Common garden design

This study is based upon the analysis of data from two common gardens initiated in separate years (2009 and 2011) and containing a total of 21 A. californica populations (Appendix S1: Table S1, Figure 1). The site for both gardens is in Newport Beach, CA (33°39’N) and within the Upper Newport Bay Ecological Preserve. Wild A. californica grows within 10 m of the garden perimeter. The site has a mean annual precipitation and temperature (from 1964-2014) of 29.9 cm and 17.6°C, respectively (Appendix S1: Table S1, Fig. 1). 

Studying plants sourced from many environments within a common garden serves as a tool for documenting the consequences of environmental displacement, an approach commonly used in forestry provenance studies (O’Brien et al., 2007). Although displacement effects can be attributed to a variety of factors (e.g., climate, soil properties, biotic communities, etc.), we interpret displacement primarily through the lens of variation in aridity for several reasons. First, the coastal sites from which we sample A. californica vary dramatically and clinally with respect to aridity (Table A1). Second, a previous study of these populations demonstrates clinal ecotypic variation in many leaf water relations traits (Pratt and Mooney 2013, Pratt et al. 2014), consistent with local adaptation to an aridity gradient. Third, genetically-based clines in leaf functional traits parallel patterns of arthropod densities along the coast (Pratt et al. 2017), suggesting a bottom-up effect of aridity on plant-quality and associated arthropod densities. We nonetheless recognize that other factors may vary latitudinally and influence plant and arthropod performance, and we discuss the implications accordingly.

The details regarding common garden construction can be found in Appendix S2, but the core design is briefly described here. For the common garden established in 2009 (hereafter the “2009 garden”), cuttings from five A. californica populations were collected along a coastal gradient in spring 2008 and grown within a greenhouse. In December 2009, the common garden was planted into three blocks each containing a pair of plots, one irrigated and the other unirrigated (Pratt and Mooney 2013, Pratt et al. 2014, 2017). While Pratt and Mooney (2013) included a precipitation manipulation that forced plants outside the precipitation that they naturally experienced in Southern California, this study focuses on the unirrigated plots experiencing an ambient Southern California climate. The plants from each source population (sample sizes ranging from 7 to 21 per population) were evenly distributed among plots and randomized within each plot. To minimize non-genetic maternal effects associated with plants cloned from cuttings (Roach and Wulff 1987), rooted cuttings were grown in the greenhouse and common garden for a total of 24 months before collecting data.

The common garden established in 2011 (hereafter the “2011 garden”) is immediately adjacent to the 2009 garden. In December 2010, we collected seed from 10 A. californica plants in each of 21 source populations, including the five populations sampled for the 2009 garden, and germinated the seed in early February 2010 in a greenhouse. In February 2011, approximately ten individuals per population (N = 210 plants total) were transplanted into a common garden and completely randomized within a 14 by 15 m grid. Plants within each garden were lightly irrigated during their first summer following transplant to increase survival. 

Climate data

We extracted and averaged 50 years (1964-2013) of monthly precipitation and temperature estimates for each population source site and the common garden from the PRISM database (PRISM Climate Group 2004; Appendix S1: Table S1). We quantify displacement specifically with respect to precipitation as a surrogate for aridity broadly because precipitation is highly correlated with both temperature (r = -0.71) and an aridity metric that incorporates temperature (e.g. Standardized Precipitation-Evapotranspiration Index [SPEI]; Thornthwaite, 1948; R²=0.99). This also enabled us to compare spatial and temporal variation in aridity through an easily interpretable common currency of precipitation. Also, although MAP includes both wet and dry season precipitation, which may have different impacts (Michalet et al. 2021), we find that variation in dry season precipitation along the coast is negligible (Appendix S1: Fig. S1). In parallel, we gathered precipitation data located < 2 km away from our common garden for 2009-2018 from a local weather station (33.67°, -117.89°) maintained by Orange County Watersheds (Appendix S1: Table S2). Because A. californica completes most of its growth during winter and spring rains (DeSimone and Zedler 2001) and we sampled arthropods in May at peak plant biomass (see below), we computed annual precipitation from October 1 to April 30 (i.e., a hydrologic year). Precipitation between May 1 and October 1 is minimal, constituting only 5% of mean annual precipitation.

Plant performance - aboveground biomass and survival

To assess the effects of climate change on aspects of plant performance relevant to arthropods, we measured plant canopy size and survival from 2010-2018 at the conclusion of each growing season (mid-May). To estimate aboveground dry biomass, we collected reference branches from an A. californica shrub outside of our garden plots and visually estimated the total number of such branches needed to reconstruct our experimental shrubs separately for two reference branches. These reference branches were then dried and weighed to estimate shrub dry biomass. Data from 2010 and 2011 in the 2009 garden were based on estimations of canopy volume (Pratt & Mooney, 2013), and we subsequently converted these volume estimates to dry biomass based upon a regression formula (F = 2063.9; P < 0.001; R²= 0.82; n = 455; biomass = 7.4*e^-4 + 0.16*e^-4*volume). At this time, we also noted plant mortality, assuming that plants first assessed as dead in May of a given year had died during the previous summer and that this was driven by precipitation in the hydrologic year preceding that summer mortality.

Arthropod abundance and composition

Each May from 2010 to 2017 we sampled arthropods from all plants. This sampling period corresponds with the end of the growing season when plant biomass and arthropod abundance were at their peak (KAM, unpublished data). To collect arthropods, we vacuumed each shrub exhaustively with an electric vacuum (3.5 HP Ridgid model #WD0970) into a fine mesh bag that was immediately placed in a cooler and transferred to a -20° freezer later that same day. Arthropods were subsequently separated from plant chaff and stored in 70% ethanol and identified to family, and morphological species within family (Oliver and Beattie 1996). Arthropod abundance was calculated as the sum of all arthropods collected from a given plant.

Because climatic change effects might impact arthropod trophic levels and feeding guilds differently (Huberty and Denno 2004, Gely et al. 2020), cascade up from herbivores to predators, or alter herbivore-predator interactions (McCluney and Sabo 2009, Barton and Ives 2014), each morphospecies was assigned to one of nine guilds based on published accounts for the taxonomic groups. The three herbivorous guilds consisted of phloem-feeding herbivores (Hemiptera, 38 morphospecies from nine families); chewing herbivores (e.g., Orthoptera, juvenile Lepidoptera); and other herbivores (i.e., pollen and nectar feeders, and adult individuals of galling species sampled by vacuum). The three remaining guilds included omnivores (mostly Hemiptera, Miridae); detritivores (e.g., Entomobryidae); and incidentals (e.g., non-feeding, adult Diptera and Hymenoptera; see (Pratt et al. 2017) for details). The four predatory guilds consisted of: web-spinning spiders (Araneae, 17 species from five families); hunting spiders (Araneae, 28 species from six families); parasitoids (Hymenoptera, 35 species from 11 families) and other predators (e.g., larval and adult Coccinellidae beetles).