Plant‐soil feedbacks in declining forests: implications for species coexistence

Abstract

Plant-soil feedbacks (PSFs) play a relevant role as drivers of species abundance, coexistence, and succession in plant communities. However, the potential contribution of PSFs to community dynamics in changing forest ecosystems affected by global change drivers is still largely unexplored. We measured the direction, strength and nature (biological vs. chemical) of PSFs experienced by coexisting tree species in two types of declining Quercus suber forests of southwestern Spain (open woodland vs. closed forest) invaded by the exotic soil pathogen Phytophthora cinnamomi. To test PSFs in a realistic community context, we focused not only on individual PSFs (i.e., comparing the growth of a tree species on conspecific vs. heterospecific soil) but also calculated net-pairwise PSFs by comparing performance of coexisting tree species on their own and each other's soils. We hypothesized that the decline and death of Q. suber would alter the direction and strength of individual and net-pairwise PSFs due to the associated changes in soil nutrients and microbial communities, with implications for recruitment dynamics and species coexistence. In support of our hypothesis, we found that the decline of Q. suber translated into substantial alterations of individual and net-pairwise PSFs, which shifted from mostly neutral to significantly positive or negative, depending on the forest type. In both cases however the identified PSFs benefited other species more than Q. suber (i.e., heterospecific positive PSF in the open woodland, conspecific negative PSF in the closed forest). Our results supported PSFs driven by changes in chemical soil properties (mainly phosphorus) and arbuscular mycorrhizal fungi, but not in pathogen abundance. Overall, our study suggests that PSFs might reinforce the loss of dominance of Q. suber in declining forests invaded by P. cinnamomi by promoting the relative performance of non-declining coexisting species. More generally, our results indicate an increase in the strength of net PSFs as natural forests become disturbed by global change drivers (e.g., invasive species), suggesting an increasingly important role of PSFs in forest community dynamics in the near future.

Toward more robust plant‐soil feedback research

“…to 15% and 40%, respectively, from the nominal rate of 5%). These ratios, like the Teste et al ratio and other recently published PSF ratios (Heinze et al 2016, Gomez-Aparicio et al 2017), lead to residual correlation because individual plants contribute to multiple data points. In our reanalysis of Teste et al, there were advantages to changing the response from biomass ratios to biomass per plant beyond helping maintain intended Type I error rates.…”

“…We repeated the Brinkman et al (2010) simulation with two other published ratios, yown;j À y foreign y foreign (e.g. These ratios, like the Teste et al ratio and other recently published PSF ratios (Heinze et al 2016, Gomez-Aparicio et al 2017, lead to residual correlation because individual plants contribute to multiple data points. to 15% and 40%, respectively, from the nominal rate of 5%).…”

“…Bezemer et al 2006, Kardol et al 2007) and y conditioned;j À y unconditioned y conditioned;j þ y unconditioned (Perkins et al 2016), and found these ratios also elevate Type I errors (i.e. It is often impossible to determine how PSF ratios were calculated based on provided descriptions (e.g., formula and stated bounds for ratio do not match [Gomez-Aparicio et al 2017, Mehrabi et al 2015, unspecified if ratio of means or mean of ratios [Stanescu and Maherali 2017], not indicated which if any terms are means [Perkins and Nowak 2013]). These ratios, like the Teste et al ratio and other recently published PSF ratios (Heinze et al 2016, Gomez-Aparicio et al 2017, lead to residual correlation because individual plants contribute to multiple data points.…”

“…to 15% and 40%, respectively, from the nominal rate of 5%). Researchers often fit one model to biomass data and a second model to biomass ratios when all the desired estimates could be obtained from the biomass (or log-biomass) model alone (e.g., Johnson et al 2016, Schittko et al 2016, Gomez-Aparicio et al 2017. In our reanalysis of Teste et al, there were advantages to changing the response from biomass ratios to biomass per plant beyond helping maintain intended Type I error rates.…”

Fire‐induced deforestation in drought‐prone Mediterranean forests: drivers and unknowns from leaves to communities

“…, Barbeta and Peñuelas , Gómez‐Aparicio et al. ) and resprouting tree species are being displaced by seeder and resprouting shrub species under drought (Peñuelas et al. ).…”