Hybrid mensurational-physiological models for Pinus taeda and Eucalyptus grandis in Uruguay. (2016)
Type of ContentTheses / Dissertations
Degree NameDoctor of Philosophy
PublisherUniversity of Canterbury
AuthorsRachid Casnati, Ana Ceciliashow all
There is a consensus that prediction systems should be complex enough to predict yield, and the effect of various combinations of forest management practices on the functioning of interactive natural systems, but at the same time maintain a low level of detail in order to have low implementation costs and facilitate their use. For this reason hybrid mensurational-physiological models have gained importance and attention, and it is expected that their adoption will increase in the near future. This study aimed to explore the potential advantages of a hybrid mensurational– physiological model compared to models currently used in forest plantation management, and provide a better understanding of their capability to improve precision and explanation maintaining a certain level of simplicity as required for forest management. This work also aimed to provide updated tools for managing Pinus taeda and Eucalyptus grandis in Uruguay. In Chapter 2, taper and volume equations were adjusted as those are essential to estimate individual volume and wood products. Emphasis was on testing compatible taper equations, since no models of this type have been developed to date for any species in Uruguay. However, variable exponent equations gave the best performance for predicting diameter at any height with the lowest prediction errors. In Chapters 3 to 5, three systems of stand level equations comprising dominant height, basal area, maximum diameter, standard deviation of diameters, and mortality were developed using differential equations through three approaches: i. Traditional time-based models using sigmoidal difference equations that restricted independent variables to age and parameters as functions of variables for region (base approach). ii. Augmented time-based models that had parameters as linear functions of water holding capacity and physiographical variables such as elevation, aspect and slope. iii. Hybrid physiological-mensurational models based on cumulative light sums since time of planting, with potential radiation-use calculated by modifiers accounting for influences of temperature, vapour pressure deficit (VPD), and water balance. These modified light sums replaced time in sigmoidal growth and yield difference equations. Water holding capacity was the most significant among the surrogate variables tested in the mensurational models for both species (Chapter 3), whereas elevation was seldom significant. Sine and cosine of aspect weighted by the slope, and slope were usually included but to a greater extent to one species than the other. Gains in accuracy of the augmented approach were small compared to the base equations. When adjusting hybrid growth models (Chapter 4), combinations of radiation modifiers were selected that yielded accurate results. It was important to determine whether or not the gains in accuracy were sufficiently high to justify dropping the least representative modifiers and lose flexibility. Differences in global radiation across terrain corresponding to a variety of slopes orientations were tested to see whether or not they significantly affected growth. Radiation-use modifiers related to water balance and vapour pressure deficit (VPD) produced the highest gains in precision; however the complete formulation (including also temperature) was preferred in order to maximize the model utility. Accounting for aspect and slope when computing radiation flux did not improve precision in any of the state variables for either species. For fitting hybrid mortality models (Chapter 5), it was hypothesised that the light-use efficiency approach could better explain the process leading to mortality because it accounts for predisposing site characteristics, recurring perturbations, and aggregation of stress. Extended periods of low water stress and short periods of high water stress were specifically tested as predictors of the probability of mortality. Results suggested that increase in stress did not influence the probability of mortality for Pinus taeda. However, stress helped explain the probability of mortality for Eucalyptus grandis with a negative effect: the accumulation of mild water stress tended to decrease the probability of mortality. For P. taeda, resource availability increased growth and decreased the probability of mortality and mortality rate, but for E. grandis, higher levels of resources increased growth, probability of mortality, and mortality rate. It was hypothesized that the eucalypt species is more sensitive to factors other than water, given a potentially higher tolerance to drought episodes and resilience compared to the pine species. A comparison of the three contrasting systems in terms of precision and bias as well as their capacity to reflect growth rates changes when site conditions vary was conducted. The comparison was extended to explore possible gains in diameter structure estimates. Results showed that precision tended to increase with higher levels of information; however explanatory variables included in the components of each approach and precision gains varied with species. Any of the three systems of equations can be applied for managing forests in Uruguay, especially for projecting diameter distributions, since the three approaches provided diameter distributions of similar accuracy. Nonetheless models based on the hybrid approach were more precise, especially for E. grandis (with precision gains between 9 and 14% among state variables). Biases of the predicted variables were similar between approaches, but consistently less for estimating mortality in long intervals in the hybrid formulation. Along with precision, this approach offered higher utility.