Technical Comments

Comment on “Impacts of species richness on productivity in a large-scale subtropical forest experiment”

See allHide authors and affiliations

Science  11 Jan 2019:
Vol. 363, Issue 6423, eaav9117
DOI: 10.1126/science.aav9117

Abstract

Huang et al. (Reports, 5 October 2018, p. 80) report significant increases in forest productivity from monocultures to multispecies mixtures in subtropical China. However, their estimated productivity decrease due to a 10% tree species loss seems high. We propose that including species richness distribution of the study forests would provide more meaningful estimates of forest-scale responses.

In a broad-scale multispecies experiment in subtropical China, Huang et al. (1) reported a strong positive species richness effect on forest productivity. After 8 years of growth, the stand-scale productivity of 16 species mixtures, by annual volume increment, was 80% greater than that in average monocultures. This study adds to biodiversity-productivity research in forests, in terms of number of tree species tested, and informs potential gains in productivity from monocultures to tree mixtures. Although species’ differences in shade tolerance may result in better use of available light in mixtures, the observed species richness effect and the generally faster growth of trees in mixtures relative to those in monocultures may be attributable to soil resources.

The observed species richness effect was significant and large relative to the values reported previously (2, 3), likely because more species were involved. However, the decrease of forest productivity predicted for a 10% species loss scenario seems overestimated. Although not tested for complete species assemblages of natural forests under different climate conditions, a possible tree species richness–forest productivity relationship represents a nonlinear pattern; that is, productivity increases with species richness in a decelerating pattern (2, 4, 5). The maximum productivity reached and the levels of species richness at which curves become saturated depend on many factors (5). This species richness–productivity relationship suggests that productivity of natural forests would not drop significantly as a result of tree species loss until species richness is reduced below a critical threshold (i.e., 40 to 50%). And as reported by Huang et al., the 5-year mean contribution to net biodiversity effect was 4% by 2-species mixture, 40% by 4-species mixture, 56% by 8-species mixture, and 0% by 16-species mixture (manual interpretation from their figure 2)—a clear indication of saturation.

Following the general species richness–forest productivity pattern described above, the effects of tree species loss on natural forests with high species richness should start in the upper portion of the curve (high productivity/high richness), not in the lower portion (low productivity/low richness). When productivity changes rapidly, the use of productivity change at low species richness exaggerates the possible responses of natural forests with high species richness, if a simple arithmetic mean of the selected low species mixtures is used. With Huang et al.’s approach in estimating responses of natural forests to 10% species loss using fitted mixed-effects models, productivity change by 2016–2017 volume increment also decreases substantially from low to high species mixtures (3.8% to 1.9% from 1 to 16 species, part of the data provided by the authors), which would obviously continue beyond 16-species mixture. Huang et al.’s calculation is based on assumptions that natural forests in subtropical China are evenly distributed within the species richness range tested (1 to 16) and have similar productivity responses around the mean at 4-species mixture. Neither of these assumptions is met. In subtropical China, forests are highly diverse and most natural forests have species richness >20 in sampling areas of 400 to 1000 m2 (614), making it difficult (or at best improbable) to predict the productivity response of natural forests directly based on results of selected tree mixtures of 1 to 16 species. However, a correction to Huang et al.’s approach can be easily made to substantially improve the relevance of forest-scale prediction with experimental results by including the species richness distribution of the study forests.

In a quick literature search for natural forests in subtropical China where sampling plots were randomly chosen, instead of deliberately selected as in biodiversity studies, the mean tree species richness was 27 (614). Of the 88 plots from nine studies located below 1800 m elevation, 0% had species richness <4, 1% <10, 64% >20, and 32% >30 (Fig. 1). About 20% of the plots were within the tree species richness range (1 to 16) tested by Huang et al. A nonlinear curve fitting (exponential decrease) and extrapolation of productivity changes (3.8, 3.0, 2.5, 2.2, and 1.9%) with species mixtures (1 to 16) produces estimates for all species richness found in the 88 plots, which resulted in a weighted mean of 1.1%.

Fig. 1 Reduction in forest productivity predicted for a 10% tree species loss/distribution of China subtropical forests by tree species richness.

The productivity decrease of natural forests in subtropical China due to 10% species loss would be either 1.1% (fitted mixed-effects models) or <0.1% (5-year mean net biodiversity effect), compared with the 2.7% found by Huang et al., based on their experimental data. The net biodiversity effect, which integrates the productivity of mixtures with that of corresponding monocultures, may be more indicative of productivity change.

References

View Abstract

Subjects

Navigate This Article