Linking Plant Traits to
Ecosystem Properties & Services

NPP is a measure of the amount of new living plant biomass that is produced per unit of space over a unit of time. Often, this is measured per square metre of ground area with units of $\mathrm{g} / \mathrm{m}^{2} / \mathrm{y}$ (or hectares for forests). Total NPP is very difficult to measure because it includes the production of roots, and most researchers only measure aboveground net primary productivity (ANPP), with the reasonable assumption that there is an allometric link between aboveground and belowground plant growth. In practice, ANPP is usually measured as the difference between the amount of aboveground living biomass at the start of the growing season and the amount of aboveground living biomass at the end of the growing season before any senescence begins.

In my opinion, the best approximation for the estimation of NPP of a single plant is given in Enquist and Niklas (2001). The empirical data include herbaceous and woody dicots and monocots as well as conifers and involve over 1200 forest stands from 46 countries.

The first scaling relationship that we can use ( $\mathrm{G}=$ aboveground growth in $\mathrm{kg} /$ year, $\mathrm{M}=$ aboveground mass in kg ) is $\log _{10} G=0.214( \pm 0.016)+0.749 \log _{10} M( \pm 0.007)+\varepsilon, n=334, r^{2}=0.972$. The second scaling relationship that we can use $(\mathrm{H}=$ height in m$)$ is $\log _{10} H=2.58( \pm 0.012)+0.264( \pm 0.019) \log _{10} M+\varepsilon, n=398, r^{2}=0.995$

Rearranging the second relationship (ignoring the residuals, which are centered at zero), we have $\log _{10} M=-9.773+3.788 \log _{10} H$ and, substituting this into the first scaling relationship, we have $\log _{10} G=-7.110+2.837 \log _{10} H$. "Annual growth" of a single plant is the same as "net primary productivity per plant".

From this, it follows that $N P P \propto H^{2.837}$

The second scaling relationship of Enquist and Niklas (2001), i.e. $\log _{10} M=-9.773+3.788 \log _{10} H$, links living biomass (M) to plant height (H). After rearranging, this gives an estimate of the total aboveground mass of an adult plant (thus proportional to aboveground carbon sequestration during the time the plant is living) given its height. The Intergovernmental Panel on Climate Change default value is 0.47 g carbon per g of dry biomass. So, carbon storage (C):

The remaining ecosystem properties and services are linked to plant traits using Grime (2001) and his CSR model. Before explaining these links, it is useful to briefly summarize the CSR model, which is meant as a broad scale generalization across the entire vascular plant group that ignores the many life history details of individual species that complicate precise predictions. In this model, the growth, survival and reproductive output of a plant is controlled by three groups of factors that are external to the plant. Grime calls the first group of factors "stress", which he defines as anything external to the vegetation itself that limits the growth rate of a plant below its potential maximum. Note that this excludes the competitive effects of other plants as a form of "stress". Examples are nutrient limitations, temperatures below or above the optimum, soil pH below or above neutrality, or limitations on soil water or soil oxygen availability. Clearly, what Grime calls "stress" is simply the converse of "potential productivity". Grime calls the second group of factors "disturbance", which is any factor that causes the destruction of living biomass. Grime calls the third group of factors "competition", which is the amount by which neighbouring plants can reduce the growth and survival of a target plant below what that plant could obtain in the absence of such neighbouring plants.

This CSR model is summarised by a triangular ("tenary") ordination (Figure 1), in which the percentage importance of each of the three groups of factors ( $C, S$ and $R$ ) are reflected, noting that the total importance always sums to $100 \%$. These three axes describe the possible different environments in which a plant can exist; it is assumed that an environment that is both extremely stressful (or unproductive) and extremely disturbed (whatever biomass is produced is quickly destroyed) prevents any plant from existing. Thus, environments close to the top (C) of the CSR triangle are both highly productive (i.e. minimal "stress" and thus with abundant soil nutrients and optimal water availability and temperature) and do not experience factors that destroy living biomass (i.e. minimal disturbance and thus with low herbivory, trampling or other abiotic factors that can kill plants or parts of plants like fire, forestry or agricultural activities). Environments close to the lower right $(\mathrm{R})$ corner of the CSR triangle are also highly productive but do experience severe and repeated events that destroy living biomass (high disturbance and thus high levels of herbivory, trampling, fire, clearcutting, mowing or ploughing). Environments close to the bottom left corner (S) of the CSR triangle are very unproductive (i.e. high "stress" and thus with few soil nutrients, low water availability or very cold or very hot temperatures) but do not experience factors that destroy living biomass. Finally, the CSR model assumes that environments that are both highly stressed (very low productivity) and highly disturbed (any biomass that can be produced in rapidly destroyed) cannot maintain any vegetation cover.

Central to Grime's CSR theory is that each combination of stress, disturbance and competition selects for different correlated combinations of trait values among the plants that successfully inhabit such environments, such that plants typical of each environment share a common syndrome of traits that allow them to survive and reproduce in that environment. Such traits are called "functional" traits. Since the values of these functional traits determine the levels of stress, disturbance and competition to which a given species is best adapted, one can use these trait syndromes to place each species somewhere within the CSR triangle. Species typical of the three extremes of the triangle (the $\mathrm{C}, \mathrm{S}$ and R corners) are called "Competitors", "Ruderals" and "Stress Tolerators" respectively; species that typically occupy the various intermediate environments within the CSR triangle have various intermediate combinations of trait syndromes.

The mapping of functional traits to these three selective forces in the environment has been a subject of research for at least 50 years, as more and more traits and species have been included. However, the inclusion of many traits when placing a species within the CSR triangle causes a practical problem when extended to many species in a global comparison: existing trait databases are full of "holes" since different researchers measure different combinations of traits for their own specific purposes. Although there exist statistical methods of filling in missing trait values (data imputation), and this website does use these methods, the accuracy decreases as more and more species are included since this implies more and more "holes" to be filled. To solve this practical problem, Pierce et al. (2017) used the very large TRY database and, by using a subset of species that had already been placed on the CSR triangle using the original (larger) set of traits, they sought to find the smallest linear combination of traits that still preserved the original CSR ordination of these traits. They ended up with only three widely measured traits that, together approximated the original CSR ordination fairly well. These traits were specific leaf area (SLA, $\mathrm{mm} \mathrm{mg}^{-2}$ ), leaf dry matter content (LDMC, $\mathrm{mg} \mathrm{g}^{-1}$ ) and leaf laminar area ( $\mathrm{cm}^{2}$ ). These three traits are among the most commonly measured functional traits of plants. The resulting method of ordination was called STRATEFY.

Such syndromes of traits, when scaled to the entire plant community, also affect certain ecosystem properties.

The production of living plant biomass (NPP) results in atmospheric $\mathrm{CO}_{2}$ being removed from the air and in soil nutrients being removed from the soil. Living plant biomass represents a pool of stored carbon and nutrients. Eventually, plant tissues die; this dead plant material is usually called "litter" in the ecological literature. This litter decomposes at different rates. As decomposition progresses, the stored $\mathrm{CO}_{2}$ is returned to the atmosphere through the respiration of decomposing organisms and the stored nutrients in the litter either return to the soil or are incorporated into the bodies of decomposer organisms. At large geographic scales, climate (temperature and precipitation) are important in controlling decomposition rate but, at local scales, decomposition rates depend mostly on the physical and chemical properties of the dead plant material as well as some physical soil properties (soil pH , local soil moisture content and soil aeration) that can vary over short geographical distances.

There is a large literature of leaf and wood properties that affect the decomposition rate of litter. Leaves with a low leaf dry matter content (LDMC), high SLA, a high concentration of remaining nitrogen, and with lower concentrations of lignin have higher decomposition rates. Many empirical studies have documented this trend and chapter 6 of Garnier and Navas (2013) reviews and documents these trends. Notice that these traits are the same ones related to a compontent of NPP (relative growth rate, or RGR), but in the opposite direction: traits that increase RGR also increase litter decomposition rate and thus the rate at which carbon and nutrients are cycled from the environment, into the living plants, and out again into the environment.

The amount of carbon stored in either living plant material or in decomposing plant litter is simply the difference between the capture of atmospheric $\mathrm{CO}_{2}$ vis NPP and its release via litter decomposition. Recalcitrant soil carbon (i.e. organic carbon that is stable and does not decompose for long periods of time) is somewhat more complicated because it is affected by physical soil properties ( pH , soil aggregation and structure), and it is also determined by chemical properties of the litter, like lignin and tannin content (which are correlated with LDMC). The production of recalcitrant soil carbon is therefore linked to the rate of litter decomposition: the slower the decomposition rate of litter, the larger the proportion of this litter that is also more recalcitrant. Therefore, we can make the following qualitative prediction. My confidence in this prediction is only slightly lower than for litter decomposition rate, simply because the question of long-term sequestration via recalcitrant carbon is less-well understood, and the relationship of plant traits and such recalcitrant carbon is less-well studied.

The S end: $\mathrm{CO}_{2}$ is only slowly captured via NPP, but the plant tissues are long-lived, the eventual slow production of litter decomposes slowly. The actual amount of carbon stored in the living and dead plant material be high.

The C-end: $\mathrm{CO}_{2}$ is rapidly captured via NPP (both due to high growth efficiency and a large standing biomass), the plant tissues are relatively short lived (except of tree trunks) but living biomass is not rapidly destroyed by density-independent events (i.e. low disturbance rates). The large amount of litter produced decomposes rapidly. The actual amount of carbon stored in the living and dead plant material be high.

The R -end: $\mathrm{CO}_{2}$ is rapidly captured (due to high growth efficiency) but less than at the C -end because there is only a small standing biomass due to frequent and intense density-independent death due to disturbances. The plant tissues are relatively short lived (little or no wood production). The litter that is produced decomposes rapidly. The actual amount of carbon stored in the living and dead plant material be low.

Nutrient cycling refers to the amount of nutrients that pass from one pool (living biomass, dead biomass, soil) to another per unit time and space. The main pools (for the vegetation component) are soils (including the soil decomposer community), living plant tissues and dead plant tissues. Nitrogen is especially important because the only inputs of new nitrogen into an ecosystem are via nitrate-fixation by Rhizobium bacteria (in symbiosis with legumes), nitrate fixation via Frankia (in symbiosis with some Alders etc.) as well as some nitrate deposition from nitrate oxide deposition (fossil fuels) and nitrate and ammonium additions in agriculture. Other soil nutrients can also be added via weathering of bedrock. Nutrient loss in a local geographical location occurs when the nutrients in any of these pools is lost to another geographical location, due to leaching or removal of biomass.

The more rapidly new plant biomass is produced, the more rapidly soil nutrients are removed from the soil and stored in the living biomass. The longer the living tissues persist, the longer these nutrients remain in this biomass pool. The more rapidly the living tissues die and become litter, and the more rapidly this litter is decomposed, the more rapidly the nutrients re-enter the soil pool. Therefore, combining what we know about NPP and decomposition, we can predict how nutrient cycling and loss will vary in the CSR triangle. My scientific confidence in these qualitative predictions is strong. Note that the C end has high rates of nutrient cycling but low levels of nutrient loss. This is because the large standing biomass rapidly captures the released nutrients from decomposition. The R end has large rates of nutrient capture by biomass but also large amounts of nutrient loss because the high levels of density-independent death (high disturbance rates). This means that the rapidly released nutrients via decomposition cannot often be captured before the nutrients are leached away.

I consider that our understanding of how plant traits relate to prevention of soil erosion, and especially how this maps onto the CSR triangle, is lower that for the previous ecosystem properties. Obviously, prevention of soil erosion is dependent of a large standing biomass of plants, and on dense and deep roots. C -dominated plant communities will have the largest standing biomass, and so C -dominated communities should be best to reduce soil erosion. Clearly, erosion is strongly dependent on soil properties, especially soil texture (sand/silt/clay content), and on precipitation rates. My best guess (with only a moderate level of confidence) is that soil erosion will be low in C -dominated communities (because there is a large amount of plant biomass and densely growing plants), will be intermediate/low in S-dominated communities and will be low in R -dominated communities.