Single-row fast-growing cadaghi (Corymbia torelliana) windbreaks are as effective as multiple-row windbreaks in reducing wind. Although all studied windbreaks reduced wind and modified leeside microclimate, maximum reductions were obtained on the leeside of less porous windbreaks and when wind direction was perpendicular to the windbreak. Older windbreak trees had significantly more roots. Roots were confined to the top 50 cm of soil and had horizontal growth preference. Root number (N) was a significant variable for estimating root length density (LV). Oven-dry whole tree weight/100 m of windbreak length ranged from 802 to 20,145 kg for 2- to 20-year-old windbreaks.
Windbreaks are used worldwide to mitigate wind-related agriculture problems. In the US Great Plains, windbreaks were planted from 1935 through 1942 in response to a decade long decrease in agricultural income, a series of dust storms and economic collapse in the wheat and corn belts. The primary objective was to stabilize microenvironment in the area. The National Windbreak Program in Australia started in 1993 to study the impact of windbreaks on microclimate and crop production (Cleugh et al., 2002). Both projects were successful and showed promising results (Munns and Stoecker, 1946; Cleugh et al., 2002). Windbreaks have also been used in South America (Peri and Bloomberg, 2002) and many developing countries (Nair, 1993). Apart from primarily reducing wind, windbreaks modifed microclimate (Cleugh, 1998, 2002; Peri and Bloomberg, 2002; Sudmeyer and Scott, 2002a; Brandle et al., 2004), reduced soil erosion and conserved nutrients (Sudmeyer and Scott, 2002a; Brandle et al., 2004; Kulshreshtha and Kort, 2009), and enhanced crop growth and increased crop yield (Cleugh, 1998; Nuberg et al., 2002; Sudmeyer and Scott, 2002b; Brandle et al., 2004). Windbreaks also provided many ecosystem services and environmental benefits including carbon sequestration and biodiversity conservation (Cleugh et al., 2002; Maize et al., 2008; Kulshreshtha and Kort, 2009). They also helped manage the spread of pathogens such as citrus canker (Xanthomonas campestris pv. citri) in South America and Australia (Leite and Mohan, 1990; Gottwald and Timmer, 1995; Muraro et al., 2001). Florida is one of the leading agricultural states in the nation. About 47,500 commercial farms produced a variety of crops, primarily vegetables and citrus, on approximately 3,743,348 ha in 2008. Florida was second in vegetable production and first in citrus production in the nation in 2007. Sales of vegetables alone exceeded $1.5 billion in 2008. The overall economic impact of Florida agriculture was estimated to be about $100 billion annually (FDACS, 2008). However, Florida agriculture still faces major challenges such as freezes, tropical storms and/or hurricanes, high winds, infertile soil, and diseases. Because of its geographical location, tropical storms and hurricanes from both Gulf of Mexico and Atlantic strike Florida frequently. Soil moisture is usually low because of the sandy soil and dry seasons. Also, Florida soils lack organic matter and cation exchange capacity (McAvoy, 2007), and nutrients easily leach during intense rain. Farms are also subjected to regular high winds. As wind eroded soils in Australia had large amounts of soil nutrients (Nuberg, 1998; Sudmeyer and Scott, 2002a), farms subjected to high winds are susceptible to soil erosion and nutrient loss. Florida’s citrus industry is severely impacted by citrus canker and greening (caused by Candidatus Liberibacter asiaticus). Wind scarring is another factor that reduces the quality of Florida agricultural products for fresh market (Albrigo, 1976; Miller et al., 1990; Miller and Burns, 1992; Núñez-Elisea and Crane, 1998; Morales and Davies, 2000; Núñez-Elisea and Crane, 2000; Stover et al., 2004). Wind scar occurs from sand abrasion and rubbing of plant parts primarily on young tender fruits. Therefore, windbreaks of ryegrass (Lolium perenne) (Workman et al., 2003) and sugarcane (Saccharum spp.) are used in vegetable farms to protect crops from wind damage. Wind speeds as low as 6.7 m s-1 have been reported to cause wind scarring in citrus (Metcalf, 1937). After extensive canker spread by 2004/05 hurricanes, windbreaks were promoted to manage canker in Florida and some citrus growers are planting windbreaks to protect citrus crop from canker and wind damages. Because of the urgency for windbreaks, fast-growing trees such as cadaghi (Corymbia torelliana), eucalypts (Eucalyptus grandis and E. amplifolia) and Australian pine (Casuarina spp.) are highly preferred due to their fast growth and evergreen nature. Cadaghi and eucalypts are not regulated in Florida, and are widely used in windbreaks (Rockwood et al., 2008). Though Australian pine is regulated because of its invasiveness, an amendment of the Florida Statute in 2009 allows commercial citrus growers only in Indian River, St. Lucie, and Martin Counties, where canker is widespread, to use Australian pine for windbreaks with a special permit. Among three species of Australian pine (C. glauca, C. equisetifolia and C. cunninghamiana) in Florida, C. cunninghamiana is considered the best for windbreaks because of its performance in other citrus growing countries and its least invasive potential (Castle et al., 2008). Trees are usually planted in multiple rows in windbreaks for optimum results, but windbreaks of only fast-growing species are rarely used. Because of limited space, current tree windbreaks in Florida are mostly of single rows. There are limited studies on the performance of fast-growing species windbreaks, planted either alone or in combination with other species (see Sun and Dickinson, 1997; Peri and Bloomberg, 2002). There is also little or no information available on the widely planted cadaghi. One of the major issues in windbreak planting is windbreak-crop competition. Competition for resources significantly impacted crop growth and reduced yield near windbreaks (Sudmeyer et al., 2002a, 2002b; Woodall and Ward, 2002; Unkovich et al., 2003). Therefore, above- and belowground competition must be minimized for optimal crop production. Compared to aboveground, belowground competition management is challenging because of the opaque soil. Underground competition near the windbreaks can be intense because fast-growing species produced relatively higher root length density (root length per unit volume of soil, LV) compared to slow-growing species (Coleman, 2007). Also, non-native species were generally competitive (Callaway and Aschehoug, 2000; Lopez-Zamora et al., 2004; Collins et al., 2007; Tamang et al., 2008). To manage underground competition effectively, information such as root architecture, distribution (both horizontal and vertical) and branching pattern is important. Carbon sequestration is an indirect benefit of windbreaks. When trees and shrubs are planted, carbon sequestration potential of farms significantly increases compared to monoculture crops. Incorporating multiple species in windbreaks increases nutrient use. Therefore, mixed species windbreaks can sequester carbon more efficiently. Along with the storage in aboveground parts, more than half of the carbon sequestered by trees is stored in the soil (Montagnini and Nair, 2004). Forests have received attention lately for their capacity to reduce carbon emissions, and extensive work has been done to estimate biomass in forests (Lambert et al., 2005; Cole and Ewel, 2006; Vallet et al., 2006; Alamgir and Al-Amin, 2008; Nogueira et al., 2008), but biomass production potential of windbreaks is little known and often neglected. For accurate biomass estimation of windbreaks, separate equations need to be developed for windbreak grown trees. In summary, windbreaks of different species and configurations have demonstrated potential for mitigating various wind-related agricultural problems across the globe (Gottwald and Timmer, 1995; Cleugh et al., 2002, Peri and Bloomberg, 2002; Brandle et al., 2004). Since wind-related agricultural issues in Florida are similar to those in other parts of the globe, windbreaks could potentially be used to mitigate such problems. However, single-row windbreaks and windbreaks of fast-growing trees such as that of cadaghi must be evaluated to see if they produce the same results and benefits of other multiple-row windbreak species. Root distribution of windbreak trees also need to be studied to effectively manage underground competition.
This study was conducted to evaluate single-row cadaghi windbreaks in southern Florida. The objectives were to study (a) single-row cadaghi windbreak function and its effect on microclimate, (b) root distribution of windbreak grown cadaghi trees, and (c) biomass in various aged cadaghi windbreaks. The hypotheses of this research were: Hypothesis 1: Single-row cadaghi windbreaks reduce wind and modify microclimate. Hypothesis 2: Older cadaghi trees produce comparatively more roots and number of roots (N) is a good predictor of root length density (root length per unit volume of soil, LV).
STUDY AREA The study was conducted at C&B Farms (26°27’30”N, 80°58’46″W) near Clewiston, Florida. Single-row cadaghi (Corymbia torelliana) windbreaks were planted along primary irrigation channels, with ~1.5 m between the windbreaks and the channels. Access roads separated windbreaks and crop fields in most places, but some windbreaks were planted at the ends of fields. East-west oriented windbreaks were planted approximately every 300 m, while the distances between north-south oriented windbreaks ranged between 480-680 m. Sugarcane windbreaks were planted parallel to north-south oriented windbreaks (between two east-west oriented windbreaks) and bordered ~0.9 ha blocks. The oldest cadaghi windbreak was planted in 1988, while the rest were planted in subsequent years. Some windbreaks were established and functional while others were in the early stages of growth. To drain surface water runoff from crop fields efficiently during the rainy season, a gentle slope is maintained from the north to the south, with north end of the field at ~50 cm higher than the south. METHODS Windbreak Function and Microclimate Modification Three established windbreaks (WB) were selected for the study (Table 1, Figure 1). WB1 was east-west oriented on the northern boundary of C&B Farms. WB2 was ~300 m south of WB1. WB3, the easternmost windbreak, was oriented north-south. Porosities of WB1, and WB3 were uniform throughout the height of the windbreak. However, pruning of lower branches created some large gaps near the ground in WB2. Five windbreak sections (~45 m long) were randomly selected in each windbreak. Total height, diameter at breast height (DBH), live crown length and spacing between trees were measured for each tree in the windbreak sections (Table 1). Pictures of randomly selected windbreak sections were used to estimate windbreak porosity using Kenny’s (1987) digitizing method. In this method pictures of windbreaks are digitized and converted to black and white pictures. Black and white pixels are then counted using digitizing software and porosity is calculated as the ratio of white to total pixels of the black and white picture. Automated weather stations (AWS) were installed in November 2007 and measurements were taken until December 2008. Sugarcane windbreaks and crops in most part of the field were fully grown. There was open space only for a transect until mid-June 2008 where wind and microclimate was not influenced by sugarcane windbreaks. Therefore, AWSs were installed only in a transect perpendicular to windbreaks WB1 and WB2. After the crops were harvested and the sugarcane windbreaks were cut in June/July 2008, another transect was installed in July. AWSs were installed at 2, 4, 6, and 10H (H is the windbreak height) from WB1 on the south side of the windbreak. Another series of AWSs were installed at 2, 4, 6, and 8H from WB2 on the north side of the windbreak in the same transect. Stations at 10, 6, 4, and 2H from WB1 were approximately at 15, 23, 27, and 31H from WB2. Similarly, stations at 8, 6, 4, and 2H from WB2 were approximately at 13, 14, 15, and 16H from WB1. AWSs were also installed at 4, 8, 12, and 16H on the west side of WB3 along two transects from July to December 2008. At each station, wind speed, temperature and relative humidity (RH) were measured at a height of 2 m above the ground. Wind speed was measured using HOBO wind speed smart sensors (S-WSA-M003), and temperature/RH was measured using HOBO temperature/RH sensors (S-THA-M002). Automatic measurements were taken every 30 seconds until July and every 1-minute after that. Measurements were recorded in HOBO Micro Station Data Loggers (H21-002). A control station located at 531 m (59H) from the windbreaks also measured temperature/RH, and wind speed and direction at 2 m above the ground using wind speed and direction smart sensors (S-WCA-M003). Root Distribution Three established windbreaks (WB) were studied (Table 2, Figure 2), including WB2 and WB3. WB4 was ~300 m south of WB2. Both WB2 and WB3 had an access road (~5 m wide) on one side and an irrigation channel (~3-3.5 m wide) on the other side at ~1.5 m from the windbreak. WB4 was at the end of the crop field and had an irrigation channel on the south of the windbreak and crop field on the north. WB2 and WB4 were both 8-years old. Five windbreak sections (~45 m long) in WB2 and WB3 that were randomly selected in each windbreak (see windbreak function and microclimate modification section in methods) were used. In shorter (length) windbreak WB4, only four windbreak sections (~45 m long) were randomly selected. Each tree in a windbreak section was measured in September 2008 for total height and DBH (Table 2). Distance between trees in WB4 was also measured. Two trees of average height and DBH were randomly selected in each windbreak and trenches (T, Figure 2) 2 m long x 1 m wide x 1 m deep were dug at 2 and 4 m from each tree (on the same side of the tree parallel to the windbreaks) using a backhoe in October/November 2008 for root sampling. Trench face was smoothed using a shovel. A 1 x 1 m area was selected on a trench face perpendicular to the tree for root measurement. Each 1 x 1 m area was divided into 10 x 10 cm grids, and the number of roots (N) in each grid was counted. Roots were classified by diameter into fine roots (50 cm in DBH (3.2%). The number of trees 5-10 m tall was the highest (44.2%) followed by 15-20 m (19.8%). Trees in all windbreaks were first grouped into DBH classes. Eleven trees were then randomly selected based on the DBH distribution and destructively sampled: 1, 2, 5, 2 and 1 trees from each of the ≤10, 10-20, 20-30, 30-40, and 40-50 cm DBH classes, respectively were selected (Table 4). Relatively more sample trees were selected from DBH classes with more trees. Since larger trees were mostly in WB1, which had wide irrigation channels on either side making access limited, trees >50 cm DBH were not included in the sample. Height and DBH of sample trees were measured before felling. Trees were then cut at ground level. The crown was divided into two equal parts: upper and lower crown. Crown (including branch and leaf) weight was estimated using randomized branch sampling (RBS; Valentine et al., 1984; Gregoire et al., 1995). In RBS, the trunk as well as branches above a threshold diameter are considered branches. A segment is defined as the part of the branch between two consecutive nodes. A sequence of connected branch segments forms a path. Two paths were randomly selected in each crown section. The selection probability assigned to each branch at a node was D2.67 (where D is the diameter) divided by the sum of the D2.67 values of all branches emanating from the node. Cumulative selection probabilities were calculated for branches at each node. Then a random number was generated between zero and one using Microsoft Excel, the branch with the cumulative selection probability larger than the random number was selected, and the path continued into another segment. The process was repeated until a terminal branch with a diameter of 2.5 cm was obtained for trees with DBH >10cm. Where trees had DBH < 10 cm, the branch diameter of 2 cm was taken as terminal. The terminal branch, i.e. the sample branch, was cut off and collected. Leaves in the sample branch were excised and separate fresh weights of leaf and branch were taken in the field. The samples were transported to the lab and oven dried at 55° C for about a month until constant weights were obtained. Segments associated with epicormic shoots and branches smaller than the threshold branch size along the path were noted and collected separately. Dry weights of sample branches and leaves from each path were used to estimate tree-level oven-dry crown weights using the inflation factors obtained from the cumulative probabilities. Four crown weight estimates were obtained for each tree from each of four paths. The average of the four crown estimates was calculated for each tree to obtain an unbiased crown weight estimate per tree (Gregoire et al., 1995). Estimates from the four paths were considered independent as the interest was in tree weight prediction rather than statistical testing. To estimate trunk weights, outside bark diameter measurements were taken at the base and then every 1.5 m along the trunk until a 2.5 cm diameter was reached. A sample disk was collected from the base and the top of each 1.5 m section. Fresh weights of the discs were taken immediately in the field. The discs were brought to the lab and soaked in water for 24 hours. After 24 hours, the discs were removed from water and excess water was wiped off. Their volume was determined using the water displacement method (Ilic et al., 2000). The discs were then oven dried at 55° C for about a month until a constant dry weight was obtained. Density of the disks was estimated in kg/m3 as described in Ilic et al. (2000). Density of each 1.5 m stem section was calculated as the average of the densities of the lower and upper disk of the section. Outside bark volume of each 1.5 m trunk section was estimated using the conic frustum equation: (1) where V is the volume (m3), l the segment length (m), D1 the outside bark diameter of the lower disk (cm) and D2 the outside bark diameter of the upper disk (cm). Oven-dry weight of each 1.5 m trunk section was estimated by multiplying the section volume by the average density of the section. Total trunk dry weight was estimated as the sum of the dry weights of all the trunk sections. Trunk and crown weight was added to obtain the whole tree weight. Development of Biomass Equations Plots of crown, trunk and whole tree weights against DBH and height were nonlinear (Figures 3 and 4). Therefore, various nonlinear equations (2 to 6) were fit using the SAS procedure PROC NLIN (SAS Institute Inc., 2008). (2) (3) (4) (5) (6) where Y is the oven-dry weight of the whole tree or tree component (crown or trunk), X the predicting variable (DBH or combinations of DBH and height), and b1-b10 are model parameters to be estimated. The root mean square error (RMSE) was calculated for each model, as well as the coefficient of determination (R2). Model 2, which had the lowest RMSE, highest R2 and best fit plots was selected. Two sets of equations, one with only DBH and another with both DBH and height as predictors, were considered. Several combinations of DBH and height were considered in the model. Models that gave the lowest RMSE, highest R2 and best fit plots were selected. The best DBH-based equations were: (7) (8) (9) where Y is the oven-dry crown (C), trunk (T) and whole tree (WT) weight (kg), DBH the diameter at breast height (cm), and b11-b16 the parameters to be estimated. The best DBH- and height-based equations were: (10) (11) (12) where Y is the oven-dry crown (C), trunk (T) and whole tree (WT) weight (kg), DBH the diameter at breast height (cm), HT the total tree height (m), and b17-b22 the parameters to be estimated. For ease of fitting the models, nonlinear models were linearized through logarithmic transformations. Equations 13-18 were obtained from logarithmic transformation of equations 7-12, respectively. (13) (14) (15) (16) (17) (18) where ln is the natural logarithm, Y the oven-dry crown (C), trunk (T) and whole tree (WT) weight (kg), DBH the diameter at breast height (cm), HT the total tree height (m), and c1-c12 the parameters to be estimated. Predictions from nonlinear and logarithmic equations were compared to see if the sum of the predicted tree component weights was equal to the predicted whole tree weight and which model gave the closest estimation to the observed whole tree weight (Table 6). The sum of predicted tree component weights was similar to predicted whole tree weight only when the same combinations of predicting variables, i.e. DBH2HT, was used for crown, trunk and whole tree weight prediction in nonlinear equations. When DBH HT was used as predicting variable for trunk weight (equation 11) and DBH2HT for crown and whole tree (equations 10 and 12), the sum of predicted tree components was not similar to predicted whole tree weight. Logarithmic transformed equations (equations 16-18) also did not have the additive property. However, mean tree weight predicted from logarithmic equations (309.3 kg) was closer to the mean of the observed whole tree weights (326.4 kg). The median observed tree weight was 262 kg. The nonlinear model underpredicted tree weight when the weight was below the median, but the prediction of logarithmic model was closer to the observed value (Figure 5). This suggested that the logarithmic model was better than the nonlinear model for smaller trees. All windbreaks except WB1 were younger than 8-year-old. Therefore, logarithmic models were selected to predict the biomass. However, nonlinear model can be used for predicting biomass in older windbreak trees. Data were fitted using the logarithmic transformed equations. Because of the limited number of samples, judging the homoscedasticity of the residuals from the logarithmic transformed equation plots was difficult. Therefore, White’s test was performed in the PROC REG procedure to test the homogeneity of error variance. The PROC UNIVARIATE procedure was used to test the normality of residuals. However, constant error variance and normal residuals could not be obtained from equation 14. Therefore the equation was not used and only the equation with DBH and height (equation 17) was used for estimating trunk weight. The unsigned deviation (δ), also known as error of estimate, was also calculated for each model using the back transformed data as follows: (19) where n was the sample size. Biomass Estimation in Windbreaks The final logarithmic transformed equations were used to estimate oven-dry weights of trees in the windbreaks using tree variables from 45 m long windbreak sections. Because back transformation of estimates from logarithmic model added systematic bias leading to underestimation of weight (Baskerville, 1972), adjustments were made to final estimates by multiplying the estimates by correction factor (Sprugel, 1983). The correction factor was calculated using following equations: (20) where SEE is the standard error of estimate, yi the value of the ith dependent variable; ?i the corresponding ith predicted values and n the number of samples. Then the correction factor, CF, was calculated as: (21) Estimated oven-dry weight in five windbreaks was expressed in kg weight/100 m windbreak length. DATA ANALYSIS Windbreak Function and Microclimate Modification Hourly averages were computed from the recorded data. Readings from two transects were also averaged for each location when available. However, data recorded at 1-minute intervals on August 19, 2008 were also used to study the patterns and extent of wind reduction during tropical storm Fay. Measured wind speed, temperature, and RH on the leeside of the windbreak were divided by the open (control station) wind speed, temperature, and RH during that interval, respectively, to get relative values. Data were filtered by wind direction and used in the analysis only when the wind direction was between 0 and 180 degrees to the windbreak. Therefore, only the measurements when wind direction was between 270 and 90 degrees were considered for WB1. For WB2 and WB3, only the measurements between wind directions of 180 to 270, and 0 to 180 degrees, respectively, were considered. Separate averages were calculated for range of wind speeds and directions, and plotted to study the pattern in wind reduction during the period. Time series plots were also used to examine extent of wind reduction and patterns in microclimate modification on the leeside of the windbreaks. Weather was considered normal when the temperatures in the open were greater than 10° C. Temperatures below 10° C in the open was considered a cold front. For temporal variation in temperature and RH, times between 7AM and 7PM was considered day and between 7PM and 7AM were considered night. It was not possible to plot data from all measurement locations. Only the data from extreme locations are presented in some cases to make the time series plots legible. As sugarcane windbreaks were tall by the end of August 2008, and influenced measurements, only data up to August were used. Root Distribution Cadaghi roots were present in all trenches at a distance of 2 m from the windbreak. In trenches at 4 m from the windbreak, roots were present only in WB3. Therefore, only the trenches at 2 m from the windbreaks were used in the analysis. Because roots were present only up to 50 cm deep in all six trenches, depths below 50 cm was not included in the analysis. N, LV and root weight means were calculated for windbreak, depth and root sizes. Though WB2 and WB4 were the same age, they were considered as a factor in the model to examine root distribution under different land uses and soil environments. All analyses were done in SAS 9.2 (SAS Institute Inc., 2008). All root variables were assumed spatially correlated. Both PROC MIXED (for continuous response variables) and GLIMMIX (for categorical or continuous response variables) provide options to take into account correlation in dependent data. The PROC MIXED procedure was used to test the differences in bulk density at 2 m from three windbreaks treating windbreak (WB) and depth as fixed effects and trench as a random effect. The interaction was not significant and excluded from the model. The model was Yijk =µ + αi +βj + cik + eijk (22) where Yijk is bulk density in windbreak i, depth j, trench k; µ is the overall mean; αi, βj and cik are the effects due to windbreak i, depth j and trench k within windbreak i, respectively; and eijk the error. For testing differences in average root number exiting the trench face (NX) at 2 m in three windbreak trees, a generalized linear model was initially considered assuming Poisson data distribution (using the PROC GLIMMIX procedure). Windbreak (WB), depth and bulk density were considered the fixed effects and trench a random effect. However, since root counts in sixty percent of the 10 x 10 cm grids were zero-valued, there was significant over-dispersion detected in this model. Therefore, zero-inflated Poisson (ZIP) regression and Hurdle models in PROC NLMIXED were estimated with the same effects. Both ZIP and Hurdle models are mixture models that are widely used to model response data with excess zeros. In the ZIP model, a proportion, ?, of the responses are zero, and the remainder follow a Poisson distribution with average number of roots, ?. A logit or probit function is used to model the inflation probability, ? and a log function is used to model the mean function, ? (Ridout et al. 1998; Liu and Cela, 2008). The final ZIP model included three covariates in the logit function for ?, and a covariate in the mean function for ?. logit(πi) = p0 + p1WB + p2Depth + p3BD (23) ?i = log(µi) = q0+ q1WB (24) where WB, depth and BD are the effects of windbreak, depth and bulk density, respectively; p0, p1, p2, p3, q0, and q1 are parameters to be estimated. The probability density function of a ZIP model is given by (25) In a Hurdle model, the probability of excess zeroes, ?, is estimated using the actual proportion of zeroes in the data, and a truncated Poisson or Negative binomial distribution is used to model positive outcomes (Ridout et al., 1998; Liu and Cela, 2008). As in the ZIP model, a log function is used to model ?. The final Hurdle model included three covariates in the logit function for ?, and a covariate in the mean function for ?. logit(πi) =m0 + m 1WB + m 2Depth + m3BD (26) ?i = log(µi) = n0+ n1WB (27) where WB, Depth and BD are the effects of windbreak, depth and bulk density, respectively; m0, m1, m2, m3, n0, and n1 are parameters to be estimated. The probability density function of a Hurdle model is given by (28) Since WB is a categorical variable, dummy variables were used to define WB2 and WB3 with WB4 as the base (intercept only) in the NLMIXED procedure. To estimate cadaghi root growth direction preference (isotropy), the PROC GLIMMIX procedure (root number being a count data) was used to compare differences in average root numbers in the three faces of the 10 x 10 x 10 cm soil cubes. Separate tests were first performed on data sets from individual windbreak trees. Data from all three windbreak trees were then combined and the test was performed on the combined data set. Interaction terms were not significant and therefore were not included in the model. The model for individual and combined data sets was Yijkl =µ + αi +βj + ck + eijkl (29) where Yijkl is the number of roots exiting the faces of the soil cube l in direction i, depth j and trench k; µ is the overall mean; αi, βj, and ck are the effects due to direction i, depth j, trench k, respectively; and eijkl the error. The PROC MIXED procedure was used to analyze LV treating windbreak (WB) and depth as fixed effects and trench as a random effect using model 30. Yijkl =µ + αi +βj + cik + (αβ)ij + eijkl (30) where Yijkl is the LV in cube l in windbreak i, depth j and trench k within windbreak i; µ is the overall mean; αi, βj, cik and (αβ)ij are the effects due to windbreak i, depth j, trench k within windbreak i and the interaction between windbreak and depth, respectively; and eijkl the error. Number of roots exiting the three faces of the soil cube were averaged by depth to get NAVG. Data from all three windbreak trees were combined and the PROC MIXED procedure was used to test LV=2NX and LV=2NAVG relationship. Interaction was not significant and was excluded from the model. The final model was Yijkl =µ + αi +βj + ck + eijkl (31) where Yijkl was the LV in soil cube l with NX or NAVG i, depth j, trench k; µ is the overall mean; αi, βj and ck are the effects due to NX or NAVG i, depth j and trench k, respectively; and eijkl the error. The PROC MIXED procedure was used to analyze root dry weight treating windbreak (WB) and depth as fixed effects and trench as a random effect using model 32. The distribution of root dry weight was significantly non-normal and was log transformed. Interaction was also considered in the model but was not significant. Yijkl =µ + αi +βj + cik + eijkl (32) where Yijkl is the log of root dry weight in cube l in windbreak i, depth j and trench k within windbreak i; µ is the overall mean; αi, βj and cik are the effects due to windbreak i, depth j and trench k within windbreak i, respectively; and eijkl the error.
WINDBREAK FUNCTION AND MICROCLIMATE MODIFICATION
1-hour average maximum wind speed was 7.6 m/s during the study period. Average wind speed on the south side of WB1 was always lower than in the open. More wind reduction occurred when wind direction was perpendicular to the windbreak regardless of wind speed (Figures 6 and A-1). Wind speed was generally less at 4H among all locations and gradually increased up to 10H, after which wind speed decreased at16H as it approached WB2. Highest relative wind speed of 50% was recorded at 10H. At other locations wind reduction was more than 50%. At 2H, wind reduction was at least 77%. Wind was detected at 2H even when the open wind speed was less than 1.5 m/s and wind direction was nearly perpendicular (90±15 degrees) to the windbreak.
For wind speeds greater or less than 5 m/s, more wind reduction was obtained when the wind direction was perpendicular than oblique (Figure 7). Regardless of wind speed, minimum wind speed was recorded at 2H when the direction was oblique (less than 45 and greater than 135 degrees to the windbreak) to the windbreak and at 4H when the direction nearly perpendicular (between 45 and 135 degrees to the windbreak).
On the north side of WB2, maximum wind reduction was at 6H (78%) when the wind direction was between 45 and 135 degrees to the windbreak (Figures 8 and A-2). When the direction was oblique (less than 45 and greater than 135 degrees to the windbreak), maximum reduction was at 2H (at least 72%). During the study, maximum wind speed on the north side of WB2 was only 60% (at 23H) of the open wind regardless of direction. As observed on the south side of WB1, wind speed gradually increased at locations further away from the windbreak and decreased at 31H as it approached WB1.
For wind speeds greater or less than 5 m/s, highest wind reduction was observed when the direction was between 45 and 135 degrees to the windbreak (Figure 9). Wind reduction was 33% at both 15 and 23H when wind speed was greater than 5 m/s whereas the reduction was only 29% at 23H when the direction was less than 45 and greater than 135 degrees to the windbreak.
On August 19, 2008, tropical storm Fay passed through the Florida peninsula. Wind direction was East-Northeast early in the morning and gradually shifted to West by midnight. Wind direction remained between these two extremes throughout the day. Open wind speed ranged between 2.8 and 13 m/s. Cadaghi windbreaks effectively reduced wind in the protected area during the storm (Figures 10 and 11).
During the tropical storm, the average wind speeds on the north side of WB2 and west of WB3 were less than 90% and 80% of the open wind speed, respectively, when the wind direction was between 0 and 180 degrees to the windbreak (Figures 10 and 11). Average wind speed was lowest at 2H north of WB2 and at 4H west of WB3. Due to WB1 at about 33H from WB2 and the sugarcane windbreak at about 20H from WB3, wind speeds at 31H from WB2 and 16H from WB3 were lower than at preceding locations.
Compared to the open, wind speed on the north side of WB2 was comparatively low most of the time with frequently higher wind speed (Figure 12). 2H and 6H had the lowest wind speeds when the direction was less than 45 and greater than 135 degrees, and between 45 and 135 degrees to the windbreak, respectively. Because of the large pores at the base of WB2, jetting effect was also observed in the protected area (Figure A-3). As the wind direction was usually perpendicular to the windbreak when wind speed was 7-9 m/s (Figure 12), wind speeds at all locations except 6H and 10H were at least 5% higher than in the open. However, when open wind was greater than 9 m/s (range: 9-13 m/s) and wind direction between 45 and 135 degrees to the windbreak, average wind speed was at least 10% lower at 23H than in the open.
On the west side of WB3, wind speeds at 4, 8, 12, and 16H were usually less than the open most of the time when the direction was between 0 and 180 degrees to the windbreak (Figure 13). Regardless of wind speed and direction, highest wind reduction occurred at 4H (at least 20%) and least at 12H (Figure 13 and A-4). When wind direction was oblique (less than 45 and greater than 135 degrees to the windbreak), wind speed at 12H was 10% higher than in the open. This was primarily because the wind direction was mostly between 160 and 178 degrees to the windbreak (~ parallel direction). When open wind speed was 12.3 m/s, maximum wind of 6.2 m/s was recorded at 12H, regardless of wind direction.
Change in temperature on the leeside of the windbreaks was less compared to wind speed. Nighttime temperatures at locations (2H and 16H) south of WB1 did not show any specific trend until the end of June, after which the measurements were usually lower than in the open (Figure A-5). Average temperatures at locations further away from the windbreak (10H, 14H and 16H) were usually higher than in the open and slightly greater than at 2H and 6H (Figure 14). When the open wind speed was less than 2.5 m/s, relative temperature at 2H was similar to open and gradually increased up to 16H where it was 2% higher. When the open wind speed was greater than 2.5 m/s, relative temperature at 2H was approximately 2% and at 16H was 0.5% lower than in the open. Variation in nighttime temperature at the same location was higher until May and was less in later months (Figure A-5). Similar patterns were observed on the north side of WB2 when the wind was from south, except that the temperatures at all locations were slightly higher compared to the open (not shown here).
Daytime temperatures at locations south of WB1 were also higher than in the open (Figure 15). During the study, temperatures at 6H from WB1 were up to 2.9 °C higher than in the open. On average relative temperatures at 6H were 2% higher, where as it was approximately 1% higher at 10, 14 and 16H.
During cold fronts, temperatures at 2 and 6H south of WB1 were up to 1.7 and 1.9° C lower than in the open, respectively (Figure 16). Temperatures at all locations were either equal or similar to the open when the open wind speed was greater than 2 m/s.
Relative Humidity (RH) Modification
South of WB1, 6H had the highest (3.8%) nighttime RH, and 16H had the lowest among all locations until the end of May when the wind direction was from the north (Figures 17 and A-6). RHs at all measurement locations were higher than in the open except at 16H, which was approximately 1% lower. Compared to RH when wind speed was 5-7 m/s, RHs at all locations (except 16H) were higher than in the open when the wind speed was below 5 m/s. When the wind direction was from the south, similar observations were made at measurement locations north of WB2. The highest RH was recorded at 6H (from WB2) among all locations (not shown here).
Daytime RH at all locations except 6H south of WB1 was higher compared to the open when the wind direction was from the north (Figure A-7). On average, RHs at 14H was similar to open and other locations had slightly lower RH than in the open, with lowest at 16H (~ 2%, Figure 18). When the wind was from the south, measurement locations north of WB2 also showed the similar pattern. 2H and 31H had slightly lower but 6H had slight higher RH than in the open (not shown here).
The soil in the area was poorly drained Myakka sand (Sandy, siliceous, hyperthermic Aeric Haplaquods). Only the top soil layer was disturbed, and the soil profile below 20 cm depth was still natural. The top soil was dark grey in color, and hardpans were present between 20-30, 30-40 and 20-30 cm depths in WB2, WB3 and WB4, respectively. White sand was generally present between 40-60 cm, and reddish brown sand was present below 60 cm. Soil bulk density at 2 m from the windbreak was highest in WB2 (Figure 19).
The effect of windbreak location (WB) on soil bulk density was significant only at a 10% level (Table 7); mean soil bulk density was significantly different between WB2 and WB4 (Table 8).
Vertical Root Distribution
The numbers of roots on the trench face were highly variable, and root distribution did not show any consistent pattern (Figure 20). Roots were present on the trench face at 4 m from the windbreak only in WB3. Roots extended beyond 3 m in WB2, but did not exit on the trench face at 4 m from the tree. All roots on the trench face at 2 m from the windbreaks were smaller than 20 mm in diameter and were restricted to the top 50 cm of soil (Figure 20).
Average numbers of roots in WB2, WB3 and WB4 trees were 4.3, 2.7 and 0.3 roots per 100 cm2, respectively. Approximately 88.2% of the roots were present in the upper 20 cm of soil in WB3, and 53.1% were present in the upper 20 cm of soil in WB2; however, all roots were restricted to the upper 20 cm in WB4 trees. WB3 and WB4 trees had relatively more medium roots (63.1% and 72.7%, respectively). Fine roots constituted 70% of the roots in WB2 trees. Coarse roots were the least prevalent in all three windbreak trees and were present only up to 20-30 cm in WB2 and WB3, and up to 10-20 cm in WB4 trees. Only WB2 and WB3 tree fine and medium roots were able to penetrate the hardpan.
Among the three models (generalized linear model using the PROC GLIMMIX procedure, ZIP model and Hurdle) that were used to estimate root number, the ZIP model was considered the best following various statistical criterion. The values of the Akaike’s Information criteria (AIC) and the small-sample bias corrected version of the AIC (AICC) were the smallest for the ZIP model (Table 9). Therefore, the results are presented only for the ZIP model (Table 10).
Mean root counts in WB2 and WB3 trees were significantly higher than in WB4 trees given that roots were present (Table 10). Depth significantly increased and bulk density decreased zero-inflation probability. Both WB2 and WB3 trees significantly decreased zero-inflation probability compared to WB4 trees.
At 4 m from the windbreak, roots were present only in WB3 trees and confined to the top 40 cm of soil (Figure 21). The average number of roots was 2.0 per 100 cm2. There were relatively more coarse roots at depths 0-10 and 10-20 cm (approximately 68% and 69%, respectively). Fine roots dominated depths 20-30 (73%) and 30-40 cm (85%). Of the total roots at 4 m, 61.2% were coarse roots.
Root distributions at 2 m from the windbreaks were anisotropic (Table 11). The frontal face of the trench wall (X) had the highest numbers of roots. In all three windbreaks individually and from all windbreaks combined, average root numbers were ranked NX > NY > NZ.
A generalized linear mixed model was used to determine the effect of direction of root growth and depth on root number. Direction of root growth had a significant effect on root number in all three windbreak trees individually (Table 12). It was also significant when tested using combined data from all three windbreak trees (Table 12).
Least square means of root number in three faces of the cube (direction) were compared. Number of roots exiting all faces were significantly different in WB2 and WB3 trees (Table 13). In WB4 trees, only the roots exiting faces X and Z were significantly different. When the data from all three windbreaks were combined (all samples), roots exiting all three faces were significantly different.
Root Length Density (LV)
Average LV was 314.9, 256.6 and 245.2 cm per 1000 cm3 of soil for WB2, WB3 and WB4 trees, respectively (Figure 22). There was no significant effect of windbreak or depth on LV, but the interaction was significant (Table 14).
The average LV of trees at 4 m from windbreak WB3 was 187.2 cm per 1000 cm3 of soil. The highest LV (40.2%) was at 10-20 cm (Figure 23). LV at other depths was similar.
Relation between LV and N
Both NX (p = 0.0046) and NAVG (p = 0.0022) were significant variables for estimating LV in cadaghi trees. The coefficient of NX was significantly different from 2 as its 95% confidence interval did not include 2. However, the coefficient of NAVG was not significantly different from 2 (Table 15).
Average root weights at 2 m from the windbreaks were 3.3, 3.4 and 1.3 g root per 1000 cm3 of soil for WB2, WB3 and WB4 trees, respectively (Figure 24). The majority of root weight (more than 79%) was in the top 20 cm of soil in WB3 and WB4 trees, while WB2 trees had most of the root weight (63.6%) between 20-40 cm. There was no significant effect of windbreak and depth on log of root weight (Table 16).
At 4 m from the windbreak, the average root weight was 1.2 g root per 1000 cm3 of soil in WB3 trees (Figure 25). Compared to other depths, the 10-20 cm depth had higher root weight.
Biomass Allocation in Tree Components
Heights and diameter at breast heights (DBH) of sample trees ranged from 5.2 to 18.2 m and 9.0 to 49.3 cm, respectively (Table 17). Almost all smaller trees were from WB3 and WB5, which had more than 44% weight on average in the tree crown, whereas larger trees from WB1, WB2 and WB4 had less than 42% weight in the tree crown on average. Whole tree weights generally increased with DBH. Crown and trunk weights averaged over all sample trees were 45.5 and 55.5%, respectively.
Allometric Biomass Equation Fitting
Equations 13-18 (except 14) were fitted using the destructively sampled data. Parameters and fit statistics from the logarithmic equations are given in Table 18. The DBH-based equation was sufficient to predict the crown weight; adding height in the equation did not change the precision of crown weight estimate. For trunk weight, acceptable residual plots were obtained only when height was incorporated into the equation. For whole tree weight estimation, adding height increased the coefficient of determination (R2) slightly, from 0.89 to 0.90.
It was difficult to clearly see the homogeneity of error variance in the residual plots because of the limited number of data points. However, the Shapiro-Wilk (W) test for normality of errors and White’s test for homogeneity of error variance gave satisfactory results for all the equations (Table 19). The absolute percentage of deviation (?) ranged between 20 and 38% (Table 18), but predicted values in logarithmic transformed models did not markedly deviate from the observed values (Figures 26-28).
Biomass in Windbreaks
Whole tree weight/100 m windbreak length ranged from 802 to 20,145 kg in WB5 and WB1 trees, respectively (Table 20). Crown weight ranged from 470 to 7,759 kg, and trunk weight ranged from 307 to 13,687 kg in WB5 and WB1, respectively. The smallest differences in crown and whole tree weight/100 m windbreak length estimated from the DBH, and DBH- and height-based equations were in WB5 trees, but WB1 trees had the largest difference both for crown and whole tree weight. The difference in crown and whole tree weights in WB5 trees was 31 and 73 kg, respectively. In WB1 trees, the difference was 409 and 1,148 kg for crown and whole tree weights, respectively.
Windbreak Function and Microclimate Modification
Wind speeds on the leeside of the windbreaks were at least 20% lower than in the open at during normal weather conditions. Wind reduction varied with windbreak porosity, wind speed and wind direction. As expected, higher reductions were observed at locations closer to windbreaks. Regardless of the wind direction, maximum wind reductions were recorded at 2H only when the direction was less than 45 degrees and greater than 135 degrees to the windbreak, and at 4H or 6H when wind direction was between 45 and 135 degrees to the windbreak. Relatively porous cadaghi windbreaks allowed more winds to pass through them when the direction was perpendicular. The extent of wind reduction on the leeside of the windbreak also depended on open wind speed and wind direction. Maximum reductions were observed at lower open wind speeds regardless of wind direction. However, for the range of same wind speeds (e.g., > 5 m/s) more wind reductions were obtained on the leeside (all windbreaks) when the wind direction was between 45 and 135 degrees to the windbreak compared to when direction was less than 45 and greater than 135 degrees to the windbreak. Average wind reductions on the leeside of the windbreak varied between ~ 20-95% during normal weather conditions, but were generally lower than in other studies. Wind speeds have been recorded between 40% and 100% on the leeside (Brenner et al., 1995; Zhang et al., 1995).
In Australia, wind reduction was generally related to windbreak porosity and wind on the leeside was equivalent to windbreak porosity (Cleugh et al., 2002). Windbreaks that had low but uniform porosity in this study were more effective in reducing wind and wind reduction was within the expected range, i.e., relative wind speeds closer to the windbreaks south of cadaghi WB1 and north of WB2 were also nearly equivalent to their porosities (Figures A-1 and A-2). Distance of wind reduction on the leeside depends on the windbreak height and extended up to 30H (Cleugh and Hughes, 2002; Vigiak et al., 2003) and 60H (Caborn, 1957). Wind reduction on the leeside of cadaghi windbreaks were observed up to 31H. Wind reduction up to longer distances at C&B Farms was also partly due to the compound effects of two windbreaks planted parallel to each other.
Reduced wind speed on the leeside influnced temperature and RH. When the open wind speed was less than 2.5 m/s, nighttime temperature south of WB1 was similar to open at 2H. Temperature gradually increased at locations further away from the windbreak and reached the maximum at 16H (~2% higher than in the open). When the wind speed was greater than 2.5 m/s, wind speed was always lower than in the open. During the day, locations closer to the windbreak had relatively higher temperatures (up to ~ 2% at 6H). Similar results were observed at locations north of WB2. Temperature generally increased with the reduction in wind speed, and the location of maximum temperature coincided with the location of minimum wind speed (Cleugh et al., 2002). Temperature and RH patterns in the current study were consistent with observations made at other locations (McAneney et al., 1990; Cleugh, 1998). Maximum temperatures were recorded at locations closer to the windbreaks where wind reduction was the most. RH was generally higher on the leeside compared to open. Foereid et al. (2002) also observed increased temperature near a willow windbreak in Denmark during the day, but a decrease at night. RH also increased in the sheltered area during the day (Sudmeyer et al., 2002a). However, compared to wind reduction, temperature and RH modifications were insignificant and such modification extended up to 10-12H on the leeside (Cleugh et al., 2002). There was also an effect of season on microclimate modificaiton in the current study. Increase in temperature on the leeside gradually decreased after May, but RH generally increased on the leeside compared to open. Plant transpiration is a cooling process that increases humidity in the surrounding area. Increases in soil moisture lowered temperature and increased RH in the later part of the study.
Results of this study suggest that windbreaks of low porosity can lower winter temperatures near the windbreak on the leeside and make freeze events worse. Temperature inversion takes place at night and stratification of air layers occurs on the leeside of the windbreak. Reduced wind on the leeside of low porous windbreaks can cause less exchange of heat between air layers. During the process, cold air, which is heavier, settles near the ground. This can lead to formation of frost in the sheltered area during calm nights (Brandle et al., 2004). In January 2009, frost formed in areas closer to cadaghi windbreak (WB1) at night during cold front. Crops near the windbreak suffered more damage compared to crops away from the windbreak (Figures B-8 and B-9), but frost did not form in areas without windbreaks. Cold sensitive crops may suffer more damage during such events. Such catastrophic events could be avoided by allowing some air to flow through the windbreak as temperature in the protected area and open seem to be similar when wind speed was greater than 2 m/s (Figure 16).
One of the conditions required for citurs canker dispersal is wind-driven rain with speeds greater than 8 m/s. At this wind speed, bacteria are dispersed within trees and from tree to tree (Timmer et al., 2000), and wind speeds greater than 8 m/s are required to force canker bacteria through stomates and wounds (Graham et al., 2004). Wind scar in citrus occurs at wind speeds as low as 6.7 m s-1 (Metcalf, 1937). Results suggest that well designed single-row cadaghi windbreaks have the potential to reduce physical damage to crops by reducing wind speed and could potentially lower canker infection in citrus.
Numbers of roots in the windbreak trees were variable and did not show any distinct distribution pattern with depth. Roots were found only in the top 50 cm of soil. Larger diameter roots were limited in the upper soil layers and only fine and medium roots were able to penetrate the hardpan. Relatively more roots were observed in the surface soil in some studies with exponential decrease with depth in some cases (Dawson et al., 2001; Laclau et al., 2001; Odhiambo et al., 2001; Sudmeyer et al., 2004; Falkiner et al., 2006).
Species characteristics and soil factors such as soil type, compaction and nutrients regulated root growth and distribution in plants (Smit et al., 2000; Lehmann, 2003). In Australia, dense soil layers were barriers for root growth. Root density decreased gradually in the upper sandy horizon, but decrease was abrupt in clayey subsoil. In some species, roots were concentrated above the clayey subsoil (Sudmeyer et al., 2004). Water and nutrient availability also impacted root density (Fabião et al., 1995; Pronk et al., 2002; Ruiz-Sanchez et al., 2005), root production (Coleman, 2007) and root distribution (Neilsen et al., 2000; Sokalska et al., 2009). Apple (Malus domestica) trees developed shallow root systems when water was readily available at the surface, but in the absence of surface water, deep roots were developed (Sokalska et al., 2009). Maritime pine (P. pinaster) also produced deep roots to access water in drier sites (Achat et al., 2008). However, cherry trees developed roots in surface soil because of the lack of competition (Dawson et al., 2001). It is likely that the hardpan, easy access to water (and potentially nutrients), and lack of competition limited root development only to the upper 50 cm of the soil.
As expected, average root number was significantly higher in the older windbreak (WB2) trees, except in WB4. Also, the number of roots in WB2 and WB3 were higher than in WB4 trees. Several factors influenced root density in WB4 trees. First, because of the gentle slope in the fields, the south end where WB4 is planted is ~50 cm lower than the north end. Water level in the primary irrigation channels at the south end is raised almost to ground level to force water in the secondary channels during cropping season from August/September to April/May. Roots generally did not grow in soils that had either seasonally (Morris and Campbell, 1991) or permanent high water table and grew up to the depth of soil above the water table (Begg et al., 1997; Falkiner et al., 2006). The presence of roots up to 40 cm in WB2, 50 cm deep in WB3 and 20 cm in WB4 trees generally corresponded to the normal water table around the windbreaks for most part of the year. Lack of aeration in the rooting zone potentially hindered root development and growth. Second, at the beginning of the cropping season (June-August), fields are disked almost up to the tree line at the south end of the field. The disks can easily penetrate 15-20 cm into the soil and can prune tree roots. As fields are disked every year, trees have to produce new roots. Almost all roots in WB4 trees were <5 mm in diameter. Despite the young age, presence of roots at 4 m in WB3 trees was surprising. Roots extended beyond 3 m in WB2 trees, but did not exit at 4 m. The road next to WB2 was used extensively while the one next to WB3 was used less frequently. Soil bulk density (both at 2 m and 4 m) in WB2 was higher than in WB3 and WB4 at all depths. Also the water in the secondary irrigation channel that runs parallel to WB3 at ~6 m from the windbreak fluctuates regularly. It is likely that the combination of less compact soil and fluctuating water (Nepstad et al., 1994) resulted in WB3 trees developing extensive root systems. Cadaghi roots were anisotropic and NX were relatively high compared to NY and NZ. However, number of roots in species such as corn (Zea mays) (Chopart and Siband, 1999) and peach palm (Bactris gasipaes) were identical in all three directions (X, Y and Z) of the soil cube. If roots are measured correctly and the theory is applied to correct field conditions, the coefficient should at least be closer to 2 when NAVG is used (Lopez-Zamora et al., 2002). But, using both NX and NAVG in several species has given inconsistent results. In slash pine (P. elliottii), the coefficient was 1 when NX from a small volume of soil was used (Escamilla et al., 1991). Young loblolly pine planted in spodosols had a relationship improved by adding root weight and depth in the model when NX was used (coefficient = 1.4) (Adegbidi et al., 2004). In other species, the coefficient was significantly different from 2 even when NAVG was used (Baldwin et al., 1971; Marriott, 1972; Lopez-Zamora et al., 2002). However, the coefficient was not significantly different from 2 in windbreak grown cadaghi trees when NAVG was used, but the coefficient was significantly different from 2 when NX was used. As small soil volume is one of the requirements for the relationship to be valid (Kendall and Moran, 1963; Melhuish and Lang, 1968), taking small soil volume may further improve this relationship. In another study, the coefficient was closer to 2 when small volume of soil was considered (Schroder et al., 1996). WB2 trees had the highest average LV of the three windbreaks. The distribution of LV with depth at 2 m from the windbreak generally corresponded with the root distribution at the same distance except for 30-40 cm depth in WB2 trees. WB4 trees having less number of roots compared to WB3 trees had the similar average LV because of the differences in the sampling depth. Though root numbers were counted in grids up to 50 cm deep in all windbreaks, sampling for LV was done to the depth of root presence, which was up to 40 cm in WB2, 50 cm in WB3 and 20 cm in WB4. LV in WB4 trees was similar at both depths whereas it was highly variable in WB2 and WB3 trees. Compared to deeper soil, other studies have observed significantly greater LV in surface soil (Peter and Lehmann, 2000; Moroni et al., 2003; Radersma and Ong, 2004; Coleman, 2007). Broadleaf species produced higher LV than pine, and roots that were relatively small in diameter had greater LV than larger diameter roots (Coleman, 2007). This has also been observed in E. nitens and E. globulus (Moroni et al., 2003). Increased root weight can be expected with increased root density (Fabião et al., 1995; Pronk et al., 2002; Ruiz-Sánchez et al., 2005). WB2 trees which had the highest average root number also had the highest root weight. Root weight was also influenced by the presence of large roots. For example, despite having either equal or less number of roots in 0-10 cm compared to other depths in all three windbreaks, root weight was comparatively higher due to the presence of relatively more large roots. Root weight decreased with depth in five agroforestry trees in India (Das and Chaturvedi, 2008), but the windbreak trees did not show any pattern. Non-native species are widely known for their competitive strength. Average root number and LV in all three windbreak trees were 10 and 8 times less, respectively, compared to naturally established 5-year-old melaleuca (Melaleuca quinquenervia) in Florida (Lopez-Zamora et al., 2004). Since water uptake (Boyer, 1985), nutrient acquisition (Nye and Tinker, 2000) and soil respiration (Luo and Zhou, 2006) largely depend on LV, windbreak grown cadaghi may be less competitive than melaleuca. However, roots of citrus trees in flatwoods soils in Florida mostly grew horizontal and were concentrated in the upper soil layer (Calvert et al., 1977; Bauer et al., 2004; Morgan et al., 2006). As cadaghi roots were restricted in the upper 50 cm of soil and showed horizontal growth preference, there is a potential for competition between citrus trees and cadaghi trees if planted closer. If cadaghi trees compete with crops, root pruning can potentially reduce competition as cadaghi roots were generally in the upper 50 cm of the soil. Biomass Estimation Larger trees generally had more trunk weight whereas smaller trees had relatively more crown weight. This was primarily due to competition between trees in the windbreak. Young trees had relatively more space to grow and they also received sunlight from all sides as they were in north-south oriented windbreaks. As the crown starts closing in, trees compete for light and growing space. Branches in the shaded area in some trees self prune because of lateral shading and crowns of trees grown in such environments tends to be narrower and shorter. Foliage was generally concentrated in the upper part of the canopy in shade grown trees (Mar:Mohler, 1947) and branchwood production was also low (Dicus and Dean, 1998). On the other hand, trees grown in sparser stands had wider and longer crowns with numerous large lower branches (Dean and Baldwin, 1996). Compared to open grown Abutilon theophrasti, Henry and Thomas (2002) also observed reduced leaf weight and leaf area in shade grown plants because of lateral shading and height of shade grown plants increased by 33%. This phenomenon was commonly observed in shade grown plants (Henry and Aarssen, 1997). As height increased, relative allocation of trunk weight also increased (Osada et al., 2004). Dossa et al. (2008) also observed relatively higher weight fractions in open-grown coffee (Coffea canephora var robusta) trees compared to shade grown coffee. One of the significant factors for higher crown weight in younger windbreak trees is also the orientation of the windbreaks. Both WB3 and WB5 (younger windbreaks) are oriented north-south whereas the older windbreaks are oriented east-west. All the destructively sampled larger trees were from east-west oriented windbreaks. Because of the orientation, trees in north-south oriented windbreaks received sunlight from all sides throughout the day. Lateral shading in north-south oriented windbreaks is minimal compared to east-west oriented windbreaks. The height to crown ratio of the trees in the windbreaks (Table 3) also suggests that north-south oriented windbreaks (WB3 and WB5) have longer crowns, which could potentially have more branches and leaves compared to east-west oriented windbreaks. Easily measured DBH and height were used to develop biomass equations to estimate tree weights in windbreaks. DBH alone gave satisfactory results for crown and whole tree weight prediction. Though height was not an important variable for crown weight estimation, its addition in the model gave better results for whole tree weight estimate. Both DBH and height were required for trunk weight prediction. DBH is the most commonly used variable in estimating trunk and whole tree weight, and is usually measured in large scale National Forest Inventories. However, others have suggested using both the DBH and height for large scale applications (Jenkins et al., 2003; Lambert et al., 2005). Therefore, both DBH-based and DBH- and height-based equations are generally developed for most species. DBH alone gave satisfactory results, but height was a secondary variable for trunk weight estimation (Lambert et al., 2005), and it brought additional information in addition to DBH (Joosten et al., 2004; Vallet et al., 2006). Including height increased the precision of estimates but this precision was obtained at the added cost of measuring height (Zhou et al., 2007). However, acceptable fit plots were obtained only when height was incorporated in the trunk biomass model indicating that the use of only DBH in trunk weight estimation of windbreak grown cadaghi may be insufficient. DBH was an essential variable for crown weight estimation but height was less applicable (Lambert et al., 2005). The current result also supports this observation. Instead of using DBH directly, stem cross sectional area (SCSA) and circumference at breast height was used for biomass estimation in agroforestry systems in some studies. In prairie windbreaks in Canada, SCSA at 1.3 m was the best predictor for aboveground weight in deciduous and coniferous species. Including height did not improve the relationship (Kort and Turnock, 1999). Stem circumference at 1.3 m and basal circumference (at 0.40 m) were the best weight predictors for Albizia adianthifolia grown as shade tree in shaded coffee agroforestry system in Togo (Dossa et al., 2008). For weight estimation of open grown juvenile tropical tree species, the number of leaves in crown, height and basal diameter has also been used as predictors (Menalled and Keltry, 2001). Fast-growing species such as cadaghi can efficiently sequester carbon compared to other windbreak species. For example, Zhou et al. (2007) estimated weights of three Russian-olive windbreaks in eastern Montana planted in single-row and double-row with Siberian peashrub (C. arborescens). The age of the windbreaks ranged between 15 and 53 years, and the estimated whole tree weight ranged between 1,744 and 4,957 kg/100 m windbreak length for 15 and 39-year-old windbreaks. Estimated whole tree weight for 53-year-old windbreak was only 3,636 kg/100 m windbreak length. However, the whole tree weight of 8-year-old cadaghi windbreak in the current study exceeds the maximum whole tree weight observed for Russian-olive tree windbreak. Kort and Turnock (1999) reported mean aboveground weight between 161.8 to 544.3 kg/tree for eight different species with ages between 33 and 53 years. The maximum weight of 544.3 kg/tree for 33-year-old hybrid poplar was still less than the average weight/tree observed in WB1 trees in the current study. CONCLUSIONS Winds on the leeside of the windbreaks were reduced and were generally lower than in the open when the direction was between 0 and 180 degrees to the windbreak. Maximum wind reductions were obtained on the leeside of low porous windbreaks. Distance and extent of wind reduction increased up to 31H between two windbreaks (WB1 and WB2) planted parallel to each other. Windbreaks also increased temperature up to 2.9° C and RH up to 20% in the protected area. However, temperature and RH modifications were relatively less compared to wind reduction. Temperature reduction near low porous windbreaks during cold fronts may be damaging to crops. Because of the seepage irrigation at C&B Farms, the water table usually remained high in the field. High water table restricted root distribution to the top 50 cm of the soil. Older windbreak trees had significantly more roots. LV and root weight distribution with depth were similar to that of root number. Cadaghi roots were anisotropic and showed horizontal growth preference. Both NX and NAVG were significant variables for estimating LV. Assuming all roots in the grids were counted, NX and NAVG can be useful to estimate LV in cadaghi. These results should be helpful to manage underground competition at the windbreak-crop interface where cadaghi trees are planted in field situations similar to C&B Farms. Both DBH and height were useful in estimating weights of cadaghi windbreak trees. Though DBH alone was sufficient for estimating crown weight, adding height gave better results for whole tree weight. Height was an important variable for estimating trunk weight. Fast-growing cadaghi windbreaks can sequester significantly more carbon than other species while providing wind speed reduction and microclimate modification at the same time. Such systems can provide higher returns to landowners if carbon credits can be traded.
Educational & Outreach Activities
Four project-related extension publications were published and are available online. This SARE study was a major part of a PhD dissertation which will be online at the University of Florida. The three major chapters of the dissertation will also be separate publications. The first one has already been submitted to Agroforestry Systems and is under review; the rest of the chapters will be submitted later.
Tamang, B., Andreu, M.G., Friedman, M.H., Rockwood, D.L., 2009. Management of field windbreaks. Electronic Data Information Source, School of Forest Resources and Conservation, University of Florida, Gainesville, FL, http://edis.ifas.ufl.edu/fr290
Tamang, B., Andreu, M.G., Friedman, M.H., Rockwood, D.L., 2009. Windbreak designs and planting for Florida agricultural fields. Electronic Data Information Source, School of Forest Resources and Conservation, University of Florida, Gainesville, FL, http://edis.ifas.ufl.edu/fr289
Andreu, M.G., Tamang, B., Rockwood, D.L., Friedman, M.H., 2009. Potential woody species and species attributes for windbreaks in Florida. Electronic Data Information Source, School of Forest Resources and Conservation, University of Florida, Gainesville, FL, http://edis.ifas.ufl.edu/fr286
Andreu, M.G., Tamang, B., Friedman, M.H., Rockwood, D., 2008. The benefits of windbreaks for Florida growers. Electronic Data Information Source, School of Forest Resources and Conservation, University of Florida, Gainesville, http://edis.ifas.ufl.edu/fr253
Tamang, B., 2009. Tree windbreak function, root distribution and biomass production in Florida. School of Forest Resources and Conservation, University of Florida, Gainesville, FL.
Tamang, B., Andreu, M.G., Rockwood, D.L., in review. Effects of single-row tree windbreaks on wind speed and microclimate in the protected area during different weather conditions in Florida farms. Agroforestry Systems.
Castle, B., Tamang, B., 2010. How the winds blow: examples of successful windbreak. Florida Citrus Show, January 27-28, Ft. Pierce, Florida.
Tamang, B., Rockwood, D.L., Andreu, M.G., 2009.
Microclimate modification by tree windbreaks in Florida Farms. Eleventh North American Agroforestry Conference, May 31-June 3, Columbia, Missouri.
Areas needing additional study
Extensive research has been done on wind reduction and microclimate modification by tree windbreaks. However, there are only limited studies on root distribution and biomass estimation in windbreak trees. Though the results of the current study are generally in agreement with other studies, there are still many questions that need to be addressed.
• Cadaghi is a good windbreak species, but is relatively a new species for Florida. Available information suggests that cadaghi is cold intolerant. Therefore, it may not be a suitable species for parts of Florida, where temperatures fall below freezing point in winter. Newly planted windbreaks must be closely observed and regularly monitored. Field trials should be conducted to confirm its suitability for other parts of Florida.
• Non-native species are usually highly competitive. However, this study and others (Sun and Dickinson, 1997; Nissen et al., 1999) provide limited information on cadaghi root distribution and its competitive strength. The results of this study may not be widely applicable because of the lack of site replications and unique field conditions such as location of trees along irrigation channels and high water table. Trees planted in different soil types and environments should be studied to get wider understanding of its root architecture and distribution.
• Cadaghi roots showed horizontal growth preference and roots were restricted in the top 50 cm of the soil. Cadaghi can potentially compete with crops in flatwoods soil where roots are concentrated in the upper soil layer (e.g. citrus). Therefore, its competitive potential should be studied.
• One of the requirements for windbreak trees is the ability to withstand high winds. This is very important for windbreaks in Florida as the state experiences frequent tropical storms and hurricanes. Cadaghi trees in the study area have suffered minimal damage during previous storms, but the current study does not identify why they are windfirm and warrants further investigation. Eucalypts trees generally had higher root density under the stump (Laclau et al., 2001; Bouillet et al., 2002). As cadaghi is closely related to eucalypts, new research should focus in areas closer to the stump.
• Biomass equations presented here are developed from 11 destructively sampled cadaghi trees, and their wider application is limited. However, this is a good first approximation given that these trees have a limited planting distribution in Florida. Cadaghi has now been planted in other areas in central and south Florida. In the future, more samples from a larger geographical area will be needed to fully calibrate these equations.
• Though cadaghi is fast-growing, wood density is relatively high compared to other fast-growing species. Cadaghi trees can produce more carbon in short period. Because of the high wood density, cadaghi may have wider applications. Cadaghi also had some medicinal value (Adeniyi et al., 2006) and was suitable for pulpwood (Guha et al., 1970). Therefore, its potential applications and markets should be explored so that the growers could be compensated for the land occupied by the windbreak either from wood products or carbon markets.
• People usually are reluctant to introduce non-native species because of invasiveness issues. Cadaghi’s potential for invading natural areas is still unknown. Seeds were dispersed by gravity and bees (Wallace and Trueman, 1995; Wallace et al., 2008) and regeneration has been observed in the understory of the current windbreaks. Its invasive potential should therefore be studied.