Recent Question/Assignment

-topic 3
Fire, kruger national park,
Every year in Kruger National Park, they fire the vegetation to help regrow new vegetation and maintain diversity. Investigation of Effect of fire intensity, fire frequency, rainfall, on the vegetation of savana in South Africa.
We are going to apply advanced R statistics to the 20yrs of data and see if there is any statistical relation of carbon emission from these fires.
description/Key words- experimental burning plots kruger, full map over 1500 km, diff precipitation gradient, rains more, more humid zones, reflects on grass biomass, trees have kanopy high (u need sufficient dry mass in ground to burn, pine is easy to burn), long high grass, we have a dataset of 20 yrs. They have split block design (aris’s paper, fire acting .. .. ) each block 7 hectares, they manipulate with seasonality (winter fire, B1 means burn eery year, B2 every 2 yrs and so on) monthly and seasonally, burning frequency and grass biomass (main agent of fire), many animals live there (lion, elephant, hayena etc for grass, 17 diff spp of deer), we are only sampling in the exp blocks which are there for last 40 yrs, and the burning pattern is consistent.
what is the effect on
which fire regime produces more fire intensity, eg fire per 6yrs should be more intense than fire every 2yrs but consider seasonality (there might be thunder, rain which might decrease intensity)
is summer annual fire regime more intense than winter for eg
we want to find out the fire intensity (kj/m/sec)
it is related to also carbon emission
paper: the effect of fire season , already done, in sense of ecology but not statistically
it doesn’t give differences between, summer annual winter annual, summer
we want find how different are those fires and link them to carbon emission
the average of year multiplied by 6
we take 6yrs (standard max interval)
paper: estimating carbon emission from African wild fires
key: fire intensity to carbon emissions
1st- produce something, run lme, inde var- plot medium fl, (meta data), fixed effect- season as.factor, fuel load (which is grass biomass), and frequency (how often they burn it) (frequency- just divide 4 by 6, or whatever it is less than 6, standard is 6)
Dependent var- plot medium fl
Indep- frequency, season, fule load
Random effect- fire year(we also have air temp humidity which is related to very correlated with year and precipitation)
Don’t need to do##Model1 -with fixed effects- run with frequency, season, grass biomass/precipitation (mdel 2, except ppt instead of grass biomass others same), see AIC, lowest ( we can also use rainfall instead of grass biomass too)
Based on what you choose, we need to nest year within landscape within plotros (random~1| year/landscape/plotros) why do we do this? Becz year is related to rainfall, humidity etc
Grass biomass and ppt is very informative. So include both in the maximal model in (fixed effects)
Run an anova
Tukey plots-(we can get them in glm or lmes) e.g graphs are 2-1 comparing burning every 2 yrs to burning every 1yr and so on, the means are very informative,
And y aixs season - bottom axis grass biomass
Start by tukey test, then lem , then plot all effects (if it doesn’t work with random effects(year/landscape/plot, remove it)
Don’t need the one with data frame.
Literature-- Heat yield for carbon emission (find the formula)
do Tacky tests to compare the differences .
run the model, do tacky plot, do model effects with all effects, calculate from to transform heat yield to carbon emission(u can use the what is the model coefficients- eg for burning every yr, 2 yrs, 3 yrs) then run a simple regression between a raw data
first find literature formula to transform heat y to carbon em( new column in excel)
multiply what u find by whatever u need to make 6. Eg. 2 so multiply by 3 to make 6,
it has to become larger than what it is to compare with 6
do another tacky plot, to see what other carbon emissions we have
refer to the theory in mixed effects for codes.
-Tacky test always compare pairwise things (so frequency
-And transform everything to 6
tukey plots R code