The diameters and heights of trees are two of the most important components in a forest inventory. In some circumstances, the heights of trees need to be estimated due to the time and cost involved in measuring them in the field. Artificial intelligence models have many advantages in modeling nonlinear height–diameter relationships of trees, which sometimes make them more useful than empirical models in estimating the heights of trees. In the present study, the heights of trees in uneven-aged and mixed stands in the high elevation forests of northern Iran were estimated using an artificial neural network (ANN) model, an adaptive neuro-fuzzy inference system (ANFIS) model, and empirical models. A systematic sampling method with a 150 × 200 m network (0.1 ha area) was employed. The diameters and heights of 516 trees were measured to support the modeling effort. Using 10 nonlinear empirical models, the ANN model, and the ANFIS model, the relationship between height as a dependent variable and diameter as an independent variable was analyzed. The results show, according to R2, relative root mean square error (RMSE), and other model evaluation criteria, that there is a greater consistency between predicted height and observed height when using artificial intelligence models (R2 = 0.78; RMSE (%) = 18.49) than when using regression analysis (R2 = 0.68; RMSE (%) = 17.69). Thus, it can be said that these models may be better than empirical models for predicting the heights of common, commercially-important trees in the study area.

Determining forest volume increment, the potential of wood production in natural forests, is a complex issue but is of fundamental importance to sustainable forest management. Determining potential volume increment through growth and yield models is necessary for proper management and future prediction of forest characteristics (diameter, height, volume, etc.). Various methods have been used to determine the productive capacity and amount of acceptable harvest in a forest, and each has advantages and disadvantages. One of these methods involves the artificial neural network techniques, which can be effective in natural resource management due to its flexibility and potentially high accuracy in prediction. This research was conducted in the Ramsar forests of the Mazandaran Province of Iran. Volume increment was estimated using both an artificial neural network and regression methods, and these were directly compared with the actual increment of 20 one-hectare permanent sample plots. A sensitivity analysis for inputs was employed to determine which had the most effect in predicting increment. The actual average annual volume increment of beech was 4.52 m3ha−1 yr−1, the increment was predicted to be 4.35 and 4.02 m3ha−1 yr−1 through the best models developed using an artificial neural network and using regression, respectively. The results showed that an estimate of increment can be predicted relatively well using the artificial neural network method, and that the artificial neural network method is able to estimate the increment with higher accuracy than traditional regression models. The sensitivity analysis showed that the standing volume at the beginning of the measurement period and the diameter of trees had the greatest impact on the variation of volume increment.

Forest ecosystems play multiple important roles in meeting the habitat needs of different organisms and providing a variety of services to humans. Biodiversity is one of the structural features in dynamic and complex forest ecosystems. One of the most challenging issues in assessing forest ecosystems is understanding the relationship between biodiversity and environmental factors. The aim of this study was to investigate the effect of biotic and abiotic factors on tree diversity of Hyrcanian forests in northern Iran. For this purpose, we analyzed tree diversity in 8 forest sites in different locations from east to west of the Caspian Sea. 15,988 trees were measured in 655 circular permanent sample plots (0.1 ha). A combination of machine learning methods was used for modeling and investigating the relationship between tree diversity and biotic and abiotic factors. Machine learning models included generalized additive models (GAMs), support vector machine (SVM), random forest (RF) and K-nearest–neighbor (KNN). To determine the most important factors related to tree diversity we used from variables such as the average diameter at breast height (DBH) in the plot, basal area in largest trees (BAL), basal area (BA), number of trees per hectare, tree species, slope, aspect and elevation. A comparison of RMSEs, relative RMSEs, and the coefficients of determination of the different methods, showed that the random forest (RF) method resulted in the best models among all those tested. Based on the results of the RF method, elevation, BA and BAL were recognized as the most influential factors defining variation of tree diversity.

The relative importance of different biotic and abiotic variables for estimating forest productivity remains unclear for many forest ecosystems around the world, and it is hypothesized that forest productivity could also be estimated by local biodiversity factors. Using a large dataset from 258 forest monitoring permanent sample plots distributed across uneven-aged and mixed forests in northern Iran, we tested the relationship between tree species diversity and forest productivity and examined whether several factors (solar radiation, topographic wetness index, wind velocity, seasonal air temperature, basal area, tree density, basal area in largest trees) had an effect on productivity. In our study, productivity was defined as the mean annual increment of the stem volume of a forest stand in m3 ha−1 year−1. Plot estimates of tree volume growth were based on averaged plot measurements of volume increment over a 9-year growing period. We investigated relationships between productivity and tree species diversity using parametric models and two artificial neural network models, namely the multilayer perceptron (MLP) and radial basis function networks. The artificial neural network (ANN) of the MLP type had good ability in prediction and estimation of productivity in our forests. With respect to species richness, Model 4, which had 10 inputs, 6 hidden layers and 1 output, had the highest R2 (0.94) and the lowest RMSE (0.75) and was selected as the best species richness predictor model. With respect to forest productivity, MLP Model 2 with 10 inputs, 12 hidden layers and 1 output had R2 and RMSE of 0.34 and 0.42, respectively, representing the best model. Both of these used a logistic function. According to a sensitivity analysis, diversity had significant and positive effects on productivity in species-rich broadleaved forests (approximately 31%), and the effects of biotic and abiotic factors were also important (29% and 40%, respectively). The artificial neural network based on the MLP was found to be superior for modeling productivity–diversity relationships.

No abstract