Closed kunstler closed 6 years ago
Link between Vcmax and Jmax The hypothesis of a coordination between Calvin-Benson cycle limited rate of assimilation and electron transport-limited rate of assimilation, leading to the assumption of a correlation between Vcmax and Jmax. Different model to link them have been proposed.
V <- 5:200
plot(V, V*1.67, type = "l", xlab = "Vcmax", ylab = "Jmax")
lines(V, exp(1.01)*V^0.89, lty = 2)
lines(V, exp(1.669)*V^0.75, lty = 3)
lines(V, exp(1.425)*V^0.837, lty = 4)
legend("topleft", legend = c("Medlyn", "Walker", "Kattge", "Wullschleger"),
lty = 1:4)
The slope of Jmax vs Vcamx (in log log) affect the light level at which photosynthesis become RuBisCO limited. A shallower slope shifts the light level to higher light value and a steeper slope to a lower light value. This will thus affect shade tolerance.
The link between Vcmax and Jmax might, however, be more complex and depend on the light level and leaf P. According to Walker et al. 2014 the effect of SLA and leaf P are however relatively small. So this is ok to use only Jmax in function of Vcmax.
Vcmax vs Narea The photosynthetic rate should scale positively with leaf Nitrogen (N) because of the large amount of N invested in RuBisCO. Several relationships have been proposed to link Vcmax and leaf N.
Sakschewski et al. 2014 Vcmax = 31.62Narea^0.801 but UNIT CONVERSION of g per m2 to kg per m2 1000^0.801
ln(Vcmax) = 3.946 + 0.921 log(N) + 0.121 log(P) + 0.281log(N)log(P), Walker et al. 2014. To get a function of only N we can take the mean leaf P (in TRY mean leaf P per area is 0.104 g m-2, in Walker et al. 2014 data this is 0.135)
Kattge et al. 2009. Vcmax = a + b * N with a different value per Biomes estimated either with Vcmax only or also constrained by Amax. Parameters from Kattge:
#With only Vcmax
a1 <- c(1.99, 6.35, 5.4, 34.05, 4.61)
b1 <- c(10.71, 25.88, 30.38, 9.71, 30.20)
# when estimation use also Amax not only Vcmax
a2 <- c(8.9, 4.19, 5.73, 6.32, 14.71)
b2 <- c(9.30, 26.19, 26.81, 18.15, 23.15)
names(a1) <- names(b1) <- names(a2) <- names(b2) <-c("Tropical_Trees_Oxisols",
"Tropical_Trees_NonOxisols",
"Temperate_Trees",
"Coniferous_Trees",
"Shrubs")
Walker et al. 2017 explore several alternative model for Vcmax:a modificytion of Walker et al. is Vcmax = exp(3.712)*N^0.650
Domingues et al. 2011 Vcmax = 10^1.57*N^0.55 # Appendix C Table A2
Narea <- seq(0.1, 10, length.out = 100)
plot(Narea, 31.62*Narea^0.801, type = "l", xlab = "Narea (g m-2)",
ylab = "Vcmax")
lines(Narea, exp(3.712)*Narea^0.650, lty = 2)
lines(Narea, 10^1.57*Narea^0.55, lty = 3)
lines(Narea, mean(a1) + Narea*mean(b1), lty = 4 , col = 2)
lines(Narea, mean(a2) + Narea*mean(b2), lty = 4 , col = 3)
legend("topleft", legend = c("Sakschewski-TRY", "Walker2017",
"Domingues2011", "Kattge2009_V_meanPFT",
"Kattge2009_VA_meanPFT"),
lty = c(1:4,4), col = c(1,1,1,2,3))
Curves for all PFT use in Kattge et al. 2009:
plot(Narea, 31.62*Narea^0.801, type = "l", lty = 0, xlab = "Narea (g m-2)",
ylab = "Vcmax")
for (i in 1:5){
lines(Narea, a1[i] + Narea*b1[i], lty = 1 , col = i+1)
lines(Narea, a2[i] + Narea*b2[i], lty = 2 , col = i+1)
}
legend("topleft", legend = names(a1),
lty = 1, col = 2:6)
Other traits such as leaf P and SLA could also influence Vcmax (Domingues et al. 2010, Walker et al. 2014). SLA have little effect on the link between Vcmax and leaf N. Leaf P have little direct effect on Vcmax but an important interaction with leaf N: at low leaf P the slope of Vcmax vs leaf N is shallow but at high leaf P this slope is steep (see Walker et al. 2014). See figure from Walker et al. 2014
It is also possible to use another approach where both Jmax is a function of traits rather than having Jmax a function of Vcmax (Domingues et al. 2010, Walker et al. 2014).
References:
So overall using the equation of Sakschewski et al. 2015 for Vcmax = f(N) seems ok as not to different from the other fitted curve and is based on all TRY data.
For Jmax = f(Vcmax) we can use either Medlyn et al. 2011 or Walker et al. 2014.
Done I added in commit c8f63c779770091368f639195faaf67e2f1d331d a new param B_lf7 for the scaling exponent of Jmax vs Vcmax set to 1 to match Medlyn. To apply the parameters of Walker et al. 2014 we need to set B_lf6 = 1.01and B_lf7 = 0.89
ok good for now
So far I'm using Sakschewski, et al. 2015 to link Vcmax to Leaf N but there is other ref. I need to write done all possibility. Then for the Link between Jmax and Vcmax I'm using Medlyn et al. 2002 (1.67) but othe roption exist as well.
same as https://github.com/traitecoevo/trait_gradients/issues/33