2016 Special Initiatives Issue No. 1 Caribbean Caribbean Foresters: A Collaborative Network forNaturalist Forest Dynamics and Regional Forestry H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott Special Issue No. 1:35–51 2016 CARIBBEAN NATURALIST
Applicability of the Design Tool for Inventory and Monitoring (DTIM) and the Explore Sample Data Tool for the Assessment of Caribbean Forest Dynamics Humfredo Marcano-Vega1,*, Andrew Lister2, Kevin A. Megown3, and Charles T. Scott2 Abstract - There is a growing need within the insular Caribbean for technical assistance in planning forest-monitoring projects and data analysis. This paper gives an overview of software tools developed by the USDA Forest Service’s National Inventory and Monitoring Applications Center and the Remote Sensing Applications Center. We discuss their applicability in the efficient planning of national forest inventories and their use for analyzing forest dynamics. By focusing on the specific targets of inquiry of the 16th Caribbean Foresters Meeting, we make use of the Design Tool for Inventory and Monitoring and the Explore Sample Data Tool to show how they can assist with completing forest-monitoring tasks like setting broad and specific objectives, selecting forest attributes of interest, assembling and evaluating existing data, setting time/cost and precision constraints, and exploring and interpreting sample data. We present a discussion of basic sampling principles used to produce estimates (and variance of estimates) of forest attributes with the goal of promoting efficient inventory planning and data analysis best practices throughout the Caribbean region.
Introduction Reliable systems to monitor the status of forest resources and to manage and analyze data are lacking for most Caribbean islands, and many countries in the region have recognized their need to develop local capacity in forest monitoring (Brown et al. 2007, Geoghegan 2002, Lewsey et al. 2004). Given that forest inventories at the national level have not been undertaken in much of the Caribbean, baseline information is essential for biodiversity conservation, assessing anthropogenic and environmental impacts on ecosystems, and for developing management strategies, especially in light of increasing demands on resources and vulnerabilities to climate change and hydro-meteorological disasters (CANARI 2013; FAO 2001a, b; Geoghegan 2002; Lewsey et al. 2004; López-Marrero and Heartsill-Scalley 2012; Lugo et al. 2012; Madramootoo and McGill 2001; Manuel-Navarrete et al. USDA Forest Service, Southern Research Station, Forest Inventory and Analysis (FIA), International Institute of Tropical Forestry, Jardín Botánico Sur, 1201 Calle Ceiba, Río Piedras, PR 00926. 2USDA Forest Service, Northern Research Station, FIA, National Inventory and Monitoring Applications Center (NIMAC), 11 Campus Boulevard, Suite 200, Newtown Square, PA 19073 USA. 3USDA Forest Service, Remote Sensing Applications Center (RSAC), 2222 West 2300 South, Salt Lake City, UT 84119 USA. *Corresponding author -
[email protected]. 1
Manuscript Editor: Tamara Heartsill Scalley 35
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
2007; Taylor et al. 2012). Consequently, there is a growing need within the insular Caribbean for technical assistance in planning forest-monitoring projects and data analysis. The substantial dependence of the region’s livelihoods on natural resources and ecosystem services coupled with staffing and budgetary constraints within forestry organizations compels the sharing of reliable technical resources (Brown et al. 2007, CANARI 2011, Leotaud and Bobb-Prescott 2013). With the intent of participating in collaborative efforts for technology-transfer processes within the Caribbean region, this paper gives an overview of software tools developed by the USDA Forest Service’s National Inventory and Monitoring Applications Center (NIMAC; Scott 2009a, Scott et al. 2009) and the Remote Sensing Applications Center (RSAC). We show the applicability of the Design Tool for Inventory and Monitoring (DTIM) and the Explore Sample Data Tool (ESDT) for forest-monitoring components such as setting broad and specific objectives, selecting forest attributes of interest, assembling and evaluating existing data, setting time/cost and precision constraints, and exploring and interpreting sample data. Our main objectives are to emphasize the importance of expectation management and planning before collecting data, and to examine the application of these tools for the assessment of long-term forest dynamics. We focus on the specific targets of inquiry of the 16th Caribbean Foresters Meeting held in the Dominican Republic in August 2013, and discuss basic sampling principles and statistical measures and the use of DTIM and ESDT to promote efficient inventory planning and data analysis best practices throughout the Caribbean region. Preliminary sampling principles and terminology Determination of the total volume of trees can be one of the many objectives of a national forest assessment. To create a census by measuring every single tree is infeasible and impractical. To address this difficulty, it is often best to acquire a sample by getting information from a representative portion of the population of interest (the study area and the trees within it). Summaries of the sample data yield estimates of population parameters, or the true statistical properties of the population of interest (Elzinga et al. 1998, Gotelli and Ellison 2004, Ortiz and Carrera 2002a). Typical summary statistics include measures of central tendency like the sample mean; measures of spread, like the sample standard deviation (SD); and measures of precision like the standard error (SE). The ultimate goal of calculating sample statistics is to estimate the true population parameters by deriving quality (precision) indices, called sampling errors or confidence intervals (CIs), that provide the data user with an idea of the accuracy of the estimates derived from the inventory. The CI is generally presented as a range of values around an estimated mean or total, and provides an index of the precision of the estimate. The confidence level (commonly 95%) is a component of the calculation of a CI. One can interpret a given CI in the context of that confidence level: it can be seen as the likelihood that the confidence intervals generated from repeated samples of the same type will include the true population mean or total value (Elzinga et al. 1998, Grell and Lister 2009, Ortiz and Carrera 2002a). 36
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
An important factor to consider when planning an inventory and setting desired precision levels (CI parameters) is the nature of the decisions or analyses we are going to make with the information, the implications of an incorrect decision or interpretation, and the spatial scale at which we want to implement our management decisions. That is, estimates of values for subdivisions of the population will generally be less precise than for the area as a whole due to the reduced number of samples. One of the logical initial planning steps of a forest assessment is to focus on determining how many samples or observations we have to collect in order to generate statistically robust estimates. An algorithm that simplifies calculation of the required sample size and avoids the use of the finite population correction factor is used in DTIM (Thompson 2012). This algorithm yields results similar to other approaches with large sample sizes associated with national forest inventories: n = (tCV% / E%)2 where t = the Student’s t-value associated with the (1-alpha/2) percentile of the t-distribution (usually set at 2 as sample size is unknown), CV = the known or hypothesized coefficient of variation of the attribute of interest, expressed as a percentage, and E = the desired sampling error or (1-alpha)% confidence interval, expressed as a percentage (Krebs 1989, Ortiz and Carrera 2002b). This equation is derived from the normal distribution, which theoretically demonstrates that 95% of the observations lie within ±1.96 SD of the mean (van Emden 2008). The coefficient of variation (CV) is a convenient measure of relative variability resulting from dividing the SD by the mean and multiplying by 100 to express it as a percentage (Gotelli and Ellison 2004, Krebs 1989, Ortiz and Carrera 2002a). To estimate values (e.g., mean, SD) for the study area and for each attribute of interest, we can use data from previous work, collect pre-sampling data by establishing a pilot study, or make an educated guess. Data from other projects should be used with care if metadata such as plot and sample size, mean, and SD are not clearly reported (Grell and Lister 2009). If there are no previous data, or a pilot study is not feasible, an estimate can be obtained by making an educated guess from previous experience or the literature. If one has knowledge from other sources about the range of values of the attribute found on the landscape (maximum and minimum values), one can divide this range by four to obtain a rough estimate of the SD. Alternatively, the expected range of measurements can be multiplied by tabled values or conversion factors according to the size of the sample to generate an estimate of the SD (Krebs 1989). The equation shown above can also be used to estimate sample size even if the variable to be measured does not exhibit a normal, bell-shaped distribution. This is because of the Central Limit Theorem, which establishes that “as sample size increases, the means of samples drawn from a population with any shape of distribution will approach the normal distribution” (Krebs 1989). Nevertheless, if the distribution of the attribute of interest is strongly skewed, the equation should be applied cautiously, and consultation with a statistician regarding alternative sampling options is recommended (Grell and Lister 2009, Krebs 1989). 37
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
Overview and Application of Tools Both the DTIM and the ESDT are housed in Microsoft Excel spreadsheets. The purpose of DTIM is to define relationships between forest-inventory design variables and monitoring objectives. By associating the monitoring objectives with the questions to be answered in order to accomplish those objectives, DTIM can be used to develop the tables of results needed to answer the inventory questions and identify the variables or metrics to collect in the field to fill the tables. ESDT is used to examine and interpret sample data, complementing DTIM in what can be seen as a monitoring toolkit to support forest-information needs. One noteworthy aspect of DTIM is that it selects the lowest number of plots needed to achieve estimates that meet a chosen precision requirement. By helping the user assess the implications of choosing different precision requirements and computing the inventory costs, the tool enables the adjustment and optimization of sample sizes to meet information needs (Scott 2009b). DTIM also facilitates the development of a field manual with data-collection procedures by documenting the planning process of forest monitoring projects and linking field-measured metrics to specific monitoring questions and objectives. When we open the DTIM spreadsheet in Excel, there are various tabs that successively assist the user in following the monitoring steps. With the first tab, Broad Objectives, users determine desired forest conditions or outcomes and list the key monitoring questions they are seeking to answer. On each page of the program, the green cells/blocks are the ones to fill in/change and the blue cells/blocks are computed by the program. As shown in Figure 1, in the blue column of Broad Objectives, there are various pre-defined objectives that have been identified as very common in forest inventories. These include forest health, ecosystem restoration, biological diversity, wildlife habitat, and carbon sequestration, among others. As new, broad objectives are identified from evolving forest-monitoring questions, new DTIM versions are programmed to keep up-to-date with desired inventory outcomes. When we assign the Broad Objectives that most closely align to our key monitoring questions, all of the other components of DTIM (generic questions and associated metrics and variables) needed to answer our questions are highlighted. For the following discussion, we will use a monitoring project for the forests of Los Haitises National Park in the Dominican Republic as an example. We will use the DTIM and ESDT tools to focus on the specific targets of inquiry of the 16th Caribbean Foresters Meeting. The main debate addressed during the meeting was how to detect change in forest stands as a result of environmental change—particularly hurricane effects and climate change. We will choose this topic as our forest-plan main issue and, because the specific targets of inquiry were changes in: (1) stem or biomass turnover, (2) tree-growth rates, (3) structural attributes, and (4) species composition, we can ask the following questions: What is the stem and biomass turnover in the forests of Los Haitises? What are the tree-growth rates in the forests of Los Haitises? How are the forest structural attributes and species composition changing? 38
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
After establishing these Key Monitoring Questions (Fig. 1), we move on to the second planning step of setting specific objectives by assigning Broad Objectives to our key questions. For the question regarding the stem-biomass turnover in the forests of Los Haitises, we can assign the Broad Objectives “forest health”, “ecosystem restoration”, and “forest productivity”. If we select another question like, How are the forest structural attributes and species composition changing?, and we click on the Highlight Row’s Objectives button, the Broad Objectives of “biological diversity” and “forest productivity” are highlighted and shown as active, because they are the ones that we have assigned to our question #3 about changes in forest structural attributes and species composition (Fig. 1). After relating our Key Monitoring Questions to Broad Objectives, we can move to the Questions tab. Here, the screen is split. The top shows the list of Broad Objectives (from the previous Objectives tab), which have been associated with a list of Generic Questions that we called Primary and Secondary Questions (Fig. 2). The bottom of the screen shows a comprehensive list of predefined questions developed for DTIM, and known to be of interest for forest inventory and monitoring projects. Here we highlight (in blue) possible specific questions related to each of our Broad Objectives. As an example, we can look at the Broad Objective biological diversity, which we have assigned to our question about changes in forest species composition, showing the List of Primary and Secondary Questions beside it, e.g., questions # 1, 9, 10 (Fig.2). Some predefined primary questions related to biological diversity in DTIM are: (1) what is the distribution of tree species across the forested landscape? and (2) what species are increasing or decreasing in ecological importance? Both questions are related to the indicator “tree abundance”. Notice that the predefined questions in front of the ampersand are the Primary Questions and those after it are considered as Secondary Questions related to each Broad Objective. Only the Primary Questions are used to identify the variables and metrics that we need to collect in the field. The values listed can be edited according to research interests by moving a Secondary question to the Primary portion of the list or vice versa. Other specific questions can also be added if needed. Following the definition of relationships between our forest monitoring objectives and selection of the questions that we are interested in answering, we progress to the Metrics tab. This tab lists the attributes (metrics) needed to make the tables of results for answering our questions. The screen is split showing the list of selected Generic Questions and selected Metrics at the top, and the list of all Metrics at the bottom (Fig. 3). The default Metrics are listed for each Generic Question that was used on the Questions tab. The Metrics in front of the ampersand are the Primary Metrics and those after it are the Secondary Metrics. Only the Primary Metrics are used, and the values listed can be edited by moving a Primary Metric in front of the ampersand to the Secondary Metric portion after the ampersand or vice versa, providing flexibility in choosing the variables we want to measure. As shown in Figure 3, some of the Metrics that we need to sample to help us determine the distribution of tree species across the forested landscape and which tree species are increasing or decreasing in ecological importance are: area of forest land (hectares or acres), number of all live 39
Figure 1. Objectives tab/page on the Design Tool for Inventory and Monitoring (DTIM) used to set broad monitoring objectives by determining forest-plan issues or desired forest conditions and developing key monitoring questions.
2016 Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
40
Special Issue No. 1
Figure 2. Questions tab/page on the Design Tool for Inventory and Monitoring (DTIM) showing the list of broad objectives associated with the list of specific primary and secondary questions, and the list of predefined questions developed for DTIM and known to be of interest in forest inventory and monitoring projects. Highlighted (in blue) are possible specific questions related to each broad objective.
2016 Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
41
Special Issue No. 1
Figure 3. Metrics tab/page on the Design Tool for Inventory and Monitoring (DTIM) listing default attributes (metrics, highlighted in blue) needed to make the tables of results for answering monitoring questions.
2016 Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
42
Special Issue No. 1
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
trees (trees), and all live gross volume (m 3 or ft3), among others. After editing each of our Key Questions, the list of Metrics will correspond to the attributes that we want to measure in the field and present in our tables of results. We open the Area Info tab after selecting attributes (Fig. 4). In this tab, we fill in information about the actual forest area of our current inventory and monitoring project. If information is available from a pilot study or previous project in the area of interest or surrounding areas that are similar, the total number of plots that were sampled is entered so that previous data can be used to calibrate the forest inventory that we are currently planning (Fig. 4). Next, we open the Precision tab. In order for all of the combinations between objectives, questions, indicators, and metrics to appear, push the Click to Append New Tables button. Each line at the top represents one or more tables (Fig. 5). The Pick List (at the bottom of the sheet) is used to create our tables by selecting the variables that will form the rows and columns of our tables of results. Variables are divided in a list of Plot and Condition variables, Tree variables, and Others. Note that the Metrics column at the top can include multiple default Metrics to use to address a Generic Question. If desired, more than one table can be created from a line by clicking on the Copy Line button. Our first table is related to Metric #1 (first green column and row to the right), which is “area of forest land”. This metric is related to the key question regarding the stem or biomass turnover in the forests of Los Haitises, which is related to the generic question about the distribution of tree species across the forested landscape. To display the key question and generic question for a different line (table),
Figure 4. Area Info. tab/page on the Design Tool for Inventory and Monitoring (DTIM) for inputting information about the area of the project, and information on total number of plots from pilot or previous projects in order to use previous data for calibrating the current forest inventory being planned. 43
Figure 5. Fragment of the Precision tab/page on the Design Tool for Inventory and Monitoring (DTIM) showing combinations of objectives, questions, indicators, and metrics. Shown at the bottom is the Pick List for creating tables by selecting the variables that will form the rows and columns of tables of results. Each line highlighted in blue at the top represents one or more tables.
2016 Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
44
Special Issue No. 1
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
click anywhere on the line and then push the Show Questions/Pick List button. As an example, a table can be constructed by selecting plot condition variable #3 of “ecological map class” for rows, and the tree variable # 55 of “species group” for the columns (see numbers of variables on green cells indicating Row and Column; Fig. 4). By picking up row and column variables for each table, the names of the different tables of results appear on the Table Name column. Some examples of table names are “area of forest land (acres) by ecomap class and species group”, “number of all live trees (trees) by forest type”, and “tree species, and mortality of all live trees (ft3) by elevation class and understory species”, among others (Fig. 5). Via the process of associating each question with each objective, indicator, and metric, we were able to create a total of 96 tables, which can potentially be useful in answering our questions about forest dynamics in Los Haitises National Park, Dominican Republic. Because we created a total of 96 tables, the recommendation is to focus only on the tables for which the precision of results is critical to decisionmaking or analyses (Scott 2009b). The tables are thus to be seen as a potential list from which we can select those that are essential for answering the key monitoring questions. We can then assign a precision level for the table total (or an individual cell) of each selected table, which takes us to our next planning step: setting time/ cost and precision constraints for key attributes corresponding with each of the tables that we have created and selected. The goal of this planning step is to choose minimum sample sizes that will allow us to generate precision indices (sampling error or CI) that meet our requirements. DTIM is designed to apply the sample-size equation discussed previously to meet this goal. In order for DTIM to apply the equation and calculate the required sample size, we need four types of information: (1) CI half-width as a percentage (a value of 10% means that the CI will be ± 10% of the estimate), (2) Confidence level (68–95% are commonly used), (3) Estimated value for the table cell of interest (data from previous work, a pilot study, or an educated guess), and (4) Sampling error as a percent of the estimate. After filling the information in DTIM, the coefficient of variation (CV%), the required sample size for current estimates, and the resulting sampling error % are then displayed. ESDT complements DTIM by assisting in the calculation of variability measures like the SD and the SE, and determining if the sample data are normally distributed. This tool is another spreadsheet developed in Microsoft Excel with two tabs: (1) The Descriptive Stats tab helps examine and summarize sample data after researchers enter sample data into the Data Column so that summary statistical measures can be calculated (Fig. 6). The Summary Table is organized into 3 sections: (a) Initial Inspection to check for outliers or observations that are numerically distant from the rest of the data, and examine the data distribution by using the data range (minimum value, maximum value), 45
Figure 6. Fragment of the Descriptive Stats tab/page on the Explore Sample Data Tool (ESDT) for exploring and summarizing sample data.
2016 Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
46
Special Issue No. 1
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
kurtosis, and skewness values. A kurtosis value of three suggests a normal distribution, while skewness values close to zero suggest a normal distribution. Note that these are methods of examining the data’s distribution, but alone do not determine normality. Other methods designed to test normality (e.g., Shapiro-Wilk) not included in ESDT should be considered to verify that the data are normally distributed. (b) Variability Estimates to check calculations like the Mean, SD, Variance, SE, and CV, in order to describe the central tendency and variation of the sample data. (c) Report Results: Describing Precision and Variability (mean ± SD, mean ± CI). (2) The Data Distribution tab enables the analyst to visually inspect whether the sample data are normally distributed by charting nine histograms with the data tallied into different numbers of equally spaced bins or classes (Fig. 7). The program shows each histogram with the sample data in blue-colored bars and a best-fit normal distribution (based on the sample data) in magenta-colored bars. If the sample data match the best-fit data in any of the histograms, then it is likely normally distributed (Grell and Lister 2009). After estimated values and sampling errors are obtained for each of the attributes in which we have the most interest, we can decide on the number of plots that we need to achieve our desired precision. After choosing the variable that will be
Figure 7. Fragment of the Data Distribution tab/page on the Explore Sample Data Tool (ESDT) to visually inspect if the sample data are normally distributed by plotting histograms of the data next to a histogram of a normally distributed dataset with the same range. 47
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
the focus within each priority table, we can determine its precision requirements. In the Precision tab of DTIM, we can calculate the required sample size. If we choose a precision of 15% sampling error (CI half-width) with a confidence level of 90%, and the data from a previous study of 50 plots reflected a sampling error of 11% for all live tree aboveground biomass, we will need to sample a total of 76 plots to achieve our desired precision. Once all the priority tables are completed, the minimum required sample sizes will be displayed. The planning process for forest-monitoring projects involves balancing cost and precision in order to address monitoring questions. If we use the number of plots needed, DTIM can calculate the total cost of our project via a SampleSizes tab, based on an estimated cost per plot that we can input from past experiences or results of a pilot project. DTIM automatically uses the largest sample-size requirement from the tables to calculate total cost. The SampleSizes tab indicates the row (table) in the Precision tab with the highest required sample size. Analysis of costs associated with initial required sample sizes might indicate that expectations are unrealistic given available budgets. Appropriate adjustments to the precision requirements can be made in order to lower the costs of the project if necessary. DTIM follows the principle of selecting the lowest number of plots needed to meet a chosen precision requirement. It guides the user through the iterative process of trying to achieve the highest precision while staying within a fixed budget or trying to achieve the lowest cost possible for a fixed precision. By virtue of documenting the steps of our planning process, DTIM permits us to go back and redefine our questions and objectives according to what we encounter in the field when establishing our forest inventory and monitoring project. Auxiliary technical assistance on the usage of features included in DTIM can be found in the guide developed by Scott (2009b). Although natural resources and ecosystem services are essential for the sustainability of Caribbean livelihoods, up-to-date baseline information from forest inventories at the national level are still lacking for many islands of the region (Leotaud and Bobb-Prescott 2013, Taylor et al. 2012). The Global Forest Resource Assessment of 2010 reports that availability of data was low only in the Caribbean (FAO 2010). The importance of the region as a global biodiversity hotspot should stimulate new forest-monitoring projects, especially when declining trends in foreign-exchange earnings from the agricultural sector have fostered greater engagement in forestry and ecotourism within the region (FAO 2001a, b). In addition, the elevated number of travelers into and within the Caribbean region, along with a large number of imported and exported goods, poses a high risk for pest infestations in the forests of the region (Meissner et al. 2009, Newton et al. 2010). This threat increases the need for forest monitoring. Through the discussion of the Design Tool for Inventory and Monitoring and the Explore Sample Data Tool, we have tried to address the increasing need for capacity-building and adequate research that will help Caribbean nations foster long-term natural-resource monitoring and sustainability practices. Coordinated efforts for the development of research agendas related to the resilience of natural resources in 48
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
light of anthropogenic and climate-change effects are needed within the Caribbean region (López-Marrero and Heartsill-Scalley 2012, Taylor et al. 2012). We deliver training in the use of DTIM and ESDT on an as-needed basis. Contact the authors for more information.
Acknowledgments We thank Tamara Heartsill-Scalley for encouragement to present our work on the applicability of the Design Tool for Inventory and Monitoring and the Explore Sample Data Tool in the 16th Caribbean Foresters Meeting, and her commitment to developing the workshop “Assessment of long-term forest dynamics in the Caribbean” as part of the meeting’s agenda. Special gratitude to Jerry Bauer from USDA Forest Service International Cooperation, Río Pedras, PR, for his support to Humfredo Marcano-Vega in order to attend the meeting and for making the event a fruitful and great success. The comments of two anonymous reviewers improved the manuscript. Literature Cited Brown, N., T. Geoghegan, and Y. Renard. 2007. A situation analysis for the wider Caribbean. International Union for Conservation of Nature and Natural Resources (IUCN) Caribbean Initiative, Gland, Switzerland. 52 pp. Caribbean Natural Resources Institute (CANARI). 2011. Facilitating participatory natural resource management: A toolkit for Caribbean managers. Laventille, Republic of Trinidad and Tobago. 100 pp. CANARI. 2013. Supporting green economy pathways in the Caribbean through action learning. Policy Brief 14. Laventille, Republic of Trinidad and Tobago. 4 pp. Elzinga, C.L., D.W. Salzer, and J.W. Willoughby. 1998. Measuring and Monitoring Plant Populations. Bureau of Land Management, National Applied Resources Science Center, BLM/RS/ST-98/005+1730, Denver, Colorado, USA. 496 pp. Food and Agriculture Organization (FAO) of the United Nations. 2001a. Land-resources information systems in the Caribbean. Summary report and recommendations. Pp. 1–6, In R. Brinkman (Ed.). Proceedings of a Subregional Workshop held in Bridgetown, Barbados, 2–4 October 2000. FAO, Rome, Italy. FAO. 2001b. Global Forest-Resource Assessment 2000, Main report. FAO Forestry Paper 140. FAO, Rome, Italy. FAO. 2010. Global Forest-Resource Assessment 2010, Main report. FAO Forestry Paper 163. FAO, Rome, Italy. Geoghegan, T. 2002. Participatory forest management in the insular Caribbean: Current status and progress to date. Caribbean Natural Resources Institute (CANARI) Technical Report No. 310. Caribbean Natural Resources Institute, Port of Spain, Republic of Trinidad and Tobago. 29 pp. Gotelli, N.J., and A.M. Ellison. 2004. A Primer of Ecological Statistics. Sinauer Associates, Inc., Sunderland, MA, USA. Grell, A., and A. Lister. 2009. Considerations for effective sampling tutorial, Exercises 2–4. US Department of Agriculture, Forest Service, Remote Sensing and Applications Center, Salt Lake City, Utah, USA. 13 pp. Krebs, C.J. 1989. Ecological Methodology. Harper and Row, New York, NY, USA. 654 pp. 49
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
Leotaud, N., and N. Bobb-Prescott. 2013. The impacts of climate change on biodiversity and biodiversity-dependent livelihoods in the Caribbean islands. Pp. 55–61, In J-L. Vernier and M. Burac (Eds.). Direction de l’Environnement, de l’ménagement et du Logement de Martinique, Centre de recherche GEODE Caraïbe. Actes du Colloque international Biodiversité insulaire: La flore, la faune et l’homme dans les Petites Antilles, Schoelcher. EA 929, Université des Antilles, Guyane. 93 pp. Lewsey, C., G. Cid, and E. Kruse. 2004. Assessing climate-change impacts on coastal infrastructure in the Eastern Caribbean. Marine Policy 28:393–409. López-Marrero, T., and T. Heartsill-Scalley. 2012. Get up, stand up: Environmental situation, threats, and opportunities in the insular Caribbean. Caribbean Studies 40(2):3–14. Lugo, A.E., E.H. Helmer, and E. Santiago-Valentín. 2012. Caribbean landscapes and their biodiversity. Interciencia 37(9):705–710. Madramootoo, C.A., and J. McGill. 2001. An integrated approach to land and waterresources management in the Caribbean. Keynote address at the first technical meeting, Caribbean Land and Water Resources Network (CLAWRENET), Bridgetown, Barbados, 1 October 2000. Pp. 101–112, In R. Brinkman (Ed.). Proceedings of a Subregional Workshop held in Bridgetown, Barbados, 2-4 October 2000. FAO, Rome, Italy. Manuel-Navarrete, D., J.J. Gómez, and G. Gallopín. 2007. Syndromes of sustainability of development for assessing the vulnerability of coupled human–environmental systems: The case of hydrometeorological disasters in Central America and the Caribbean. Global Environmental Change 17:207–217. Meissner, H., A. Lemay, C. Bertone, K. Schwartzburg, L. Ferguson, and L. Newton. 2009. Evaluation of pathways for exotic plant-pest movement into and within the Greater Caribbean Region. Caribbean Invasive Species Working Group (CISWG), Plant Epidemiology and Risk Analysis Laboratory (PERAL), Center for Plant Health Science and Technology (CPHST), and US Department of Agriculture (USDA), Raleigh, NC, USA. Newton, L., H. Meissner, and A. Lemay. 2010. Forestry-related pathways for the movement of exotic plant pests into and within the Greater Caribbean Region. Pp. 111, In K.A. McManus and K.W. Gottschalk. (Eds.). Proceedings. 21st US Department of Agriculture interagency research forum on invasive species 2010; Annapolis, MD. General Technical Report NRS-P-75. US Department of Agriculture, Forest Service, Northern Research Station, Newtown Square, PA, USA. 156 pp. Ortiz, E., and F. Carrera. 2002a. Estadística básica para inventarios forestales. Pp. 69–98, In L. Orozco and C. Brumér. (Eds.). Inventarios Forestales para Bosques Latifoliados en América Central. Serie técnica, manual técnico no. 50, Centro Agronómico Tropical de Investigación y Enseñanza, Turrialba, Costa Rica. 264 pp. Ortiz, E., and F. Carrera. 2002b. Muestreo en inventarios forestales. Pp. 99–117, In L. Orozco and C. Brumér. (Eds.). Inventarios Forestales para Bosques Latifoliados en América Central. Serie técnica, manual técnico no. 50, Centro Agronómico Tropical de Investigación y Enseñanza, Turrialba, Costa Rica. 264 pp. Scott, C.T. 2009a. Overview of the National Inventory and Monitoring Applications Center (NIMAC). Pp. 223–229, In W. McWilliams, G. Moisen, and R. Czaplewski. (Eds.). Forest Inventory and Analysis (FIA) Symposium 2008. US Department of Agriculture, Forest Service, Rocky Mountain Research Station, Ft. Collins, Colorado. RMRS-P56CD. 839 pp. Scott, C.T. 2009b. Using the Design Tool for Inventory and Monitoring (DTIM). Available online at http://www.fs.fed.us/emc/rig/documents/DATIM/2010-09-24_Design_Tool_ for_Monitoring_Guide.pdf. Accessed 27 January 2013. 50
2016
Caribbean Naturalist H. Marcano-Vega, A. Lister, K.A. Megown, and C.T. Scott
Special Issue No. 1
Scott, C.T., R. Bush, and K. Brewer. 2009. Design Tool for Inventory and Monitoring. Pp. 230–237, In W. McWilliams, G. Moisen, and R. Czaplewski. (Eds.). Forest Inventory and Analysis (FIA) Symposium 2008. US Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, CO. RMRS-P-56CD. 839 pp. Taylor, M.A., T.S. Stephenson, A.A. Chen, and K.A. Stephenson. 2012. Climate change and the Caribbean: Review and response. Caribbean Studies 40(2):169–200. Thompson, S.K. 2012. Sampling, Third Edition. John Wiley and Sons, Inc., Hoboken, NJ, USA. Available online at doi: 10.1002/9781118162934.ch4. Accessed 27 January 2013. van Emden, H.F. 2008. Statistics for Terrified Biologists. Blackwell Publishing, Malden, MA, USA. 360 pp.
51
