MARKET STUDY

EAST EVERGLADES REPORT

FIRST REVISION

JOB# 5893DOJ

TABLE OF CONTENTS

SUMMARY OF CONCLUSIONS
VARIABLE	FINDINGS	RECOMMENDED ACTION
Project Influence	There is no evidence to suggest that either the NPS announcement of the expansion of Everglades National Park or The Corps of Engineers announcement of its changes to the L31-N canal had any measurable impact on the value of the freshwater wetlands.	An appraiser or analyst should not make any adjustment for Project Influence.
Financing	There is evidence that seller financing has an upward impact on price, even after making mathematical cost equivalency adjustments.	Owner financed transactions should be avoided, if possible, for other than agricultural use tracts. Appraisers should rely on cash or third-party financed sales for other than agricultural use sales. There is no evidence that price is affected by owner financing for agricultural use properties.
Market Conditions (Time)	Our study of many land sales in the freshwater wetlands uncovered no persuasive evidence of any change in the price per acre directly associated with the date of sale. Neither Recreational use land sales nor Agricultural land use sales changed due to changes in the date of sale, 1974-1997.	No adjustment necessary for date of sale (Market Conditions).
Property Size	There is strong evidence that "size" is a significant factor for properties generally based on size categories, especially 6-15 acres, 15-55 acres and >100 acres. There is a modest ½%/acre decrease in $/acre as a continuous variable, but this adjustment is not meaningful unless there are very large differences between the size of the comparable sales and the subject. Size is often an interaction variable. For example, a 6-15 acre tract on a paved road or a 6-15 acre tract near a bridge sold for significantly different prices than can be explained by either size or proximity.	Limit comparable sales to similar sized properties generally within the size categories identified within the study. Pay particular attention to 6-15 acre sales on paved roads, 6-15 acre tracts near a bridge and 15-55 acre tracts on dirt/grass roads as statistically significant, distinct markets.
Location	Sales within the Dade county Overlay Ordinance (within Management Areas) sold for a significantly different (lower) price than sales not governed by the Overlay Ordinance. Sales near one of the four bridges across the L31-N canal sell for substantially more than other sales. If there is no all weather road providing direct access to a parcel, being within a mile of a paved road probably affects the value.	Select comparable sales located within the Overlay Ordinance if the subject is within the area governed by the Overlay ordinance. Pay attention to the proximity of the subject or comparables to one of the bridges crossing the L-31N canal. Check for proximity to a paved road if the subject or comparables do not have either a paved or gravel/limestone road providing direct access.
Land Use Regulations	There is good evidence that this variable is a significant factor influencing land prices, specifically regarding restrictions on use imposed by the Management Area overlay zone. Properties in MA-1 and MA-3B sell for substantially more than parcels in other management areas, and parcels in MA 2 sell for less than parcels in other management areas.	There are too many other variables that affect value to directly measure the dollar impact of the numerous regulations, all other variables held constant. Select comparable sales to be relied upon from those within the overlay zone. Refer to the study to understand the relative values within each Management Area.
Road Frontage/Access	This variable has a significant influence on land prices, especially with regard to frontage on a paved road, a gravel/limestone road or a dirt/grass road. The data indicates no distinction in price between access by airboat, swamp buggy or foot paths (no vehicular access).	Try to select comparables with the same access/type of frontage as the property under appraisal. We found that properties with dirt road or grass road access sold for about 30% less per acre than p4roperties with all weather access; properties with airboat, swamp buggy or no road access sold for about 40% less per acre than properties with all-weather road access.
Mineral Rights	There is no evidence of any change in the price per acre, for sales in the freshwater wetlands, directly associated with the issue of Mineral Rights.	No adjustment necessary
Severable Use Rights	There is no evidence of any change in the price per acre, for sales in the freshwater wetlands, directly associated with the issue of Severable Use Rights.	No adjustment necessary
Compiled by Pritchett, Ball & Wise, Inc.

MARKET STUDY
EAST EVERGLADES REPORT

FIRST REVISION
03/99

DESCRIPTION OF THE FIRST REVISION

This study revision replaces the original study published 10/21/97.

Following our publication of the original East Everglades study, dated 10/21/97, the U.S. Department of Justice discovered that some of the governmental agencies involved had not sent us sales that may have reflected the impact of "project influence" on the value of land in the freshwater wetlands of south Florida. The exclusion of these sales was an understandable oversight. Agencies had been asked to provide us sales used in appraisal reports for agency land acquisition over a 20-year period. Appraisers, conscious of the Federal Land Acquisition rules, excluded sales as possible comparables if they suspected that the sales reflected "project influence". As a consequence of the exclusion of these sales from our original study, we concluded that the announcement of the East Everglades expansion in December 1989 and the announcement of the Corps of Engineers project in May 1992 had no impact on the value of the subject lands.

The Department of Justice attorneys provided us with 173 additional sales to be considered in a re-examination of our market study conclusions. Some of the additional sales were provided in an attempt to include additional agricultural use sales and sales from management areas (as described by the Dade County Overlay Ordinance discussed below) where the number of sales in the original study database was smaller than desirable. However, most of these sales were included because they may have indicated that "project influence" was a significant variable.

Inclusion of these new sales required a complete re-analysis of the data. We were obligated to geographically locate (geocode) these new sales, inspect the sales and verify the transactional information. Through this process we added a total of 130 sales to the analytical (statistical) study and 147 new sales to the descriptive database. We included sales in the descriptive database even if we learned that they did not represent arms-length transactions, because we want any analyst using the database to benefit from our research. We clearly labeled these "non-market" transactions in the descriptive database. In many circumstances we were required to undertake a lengthy verification process in order to learn that a sale actually did not represent a "market" transaction. Given the imperfect state of real estate transactional information, it is easy for an appraiser to be misled about the third-party nature of a sale. We included only what we believed to be third party, market-based transactions in the analytical study.

In the process of redoing the study, we tested for coding errors and determined that although some errors existed, they did not have a statistically significant influence on our conclusions. We corrected any coding errors that were discovered, thereby increasing the reliability of the revised study. In this second revision we more closely analyzed the potential impact of changes in the market, as reflected by date-of-sale. Our preliminary analysis indicated that although there was no evidence of change in the market 1972-1989, regardless of property type or location, but that there was an indication that sales in four of the Dade County Land Use Overlay Ordinance Management Areas (MA1, MA3A, MA3B and MA3C) increased over time, 1989-1997. Our more detailed research indicates that there is no reliable evidence that there has been a change in value over time.

We have chosen to completely replace the original study with this first revision. Our intention is to permit users to dispose of the original study and replace it with this revision without losing any of the data or information contained in the original report.

I. INTRODUCTION, PURPOSE AND FUNCTION OF THE STUDY

As a part of its program of curative measures for water flowage problems in the project areas, the United States Government is proposing to expand both the Big Cypress National Preserve and the Everglades National Park. Also, the Corps of Engineer's is engaged in a project to construct a levy to protect properties outside of the East Everglades Expansion Area from additional flooding when the water flow into the Everglades National Park is increased. The acquisitions associated with these projects will, in turn, require the services of many appraisers.

In response to a request for proposal dated April 3, 1997, Pritchett, Ball & Wise, Inc. was engaged to perform a market. This market study was performed for the Department of Justice, in cooperation with the National Park Service and the Army Corps of Engineers.

The purpose of this market study is to identify the variables that affect the value of the freshwater wetlands in south Florida. We focused on the quantitative and qualitative legal, locational and physical impacts of the likely causative (independent) variables on the market value of properties in the project area. No value conclusions were estimated for the properties and this study should not be construed as an appraisal of any single parcel. This study includes only vacant or essentially vacant land transactions and does not consider improved properties. This report is specific to the Everglades National Park only. We determined that sales associated with the Big Cypress National Preserve represented a different market area, and we prepared a separate report that is a study of variables affecting the value of the freshwater wetlands on the western side of south Florida.

This study provides (where possible and appropriate) an indication of the direction and the extent to which the independent variables appear to affect the price paid for land in this unusual and ecologically sensitive area. This study is of a scope that is much larger than is usually available to appraisers, and we were able to utilize analytical models that are more sensitive and reliable than generally available to appraisers. However, our best service to appraisers and others who utilize this report is to identify the variables that an appraiser should consider when identifying and adjusting comparable sales. The individual appraiser who is concentrating on a specific subject property is better able to judge the actual impact on price associated with the variables he or she is considering when selecting comparables and making appropriate adjustments.

Once we had accounted for interaction variables (such as size of parcel and proximity to a bridge), we developed a list of 177 potential independent variables that represented the observable factors about the sales and the market area. Only a few of the potential independent variables turned out to have any impact on the value of land in the freshwater wetlands.

The following factors were specifically identified for study in the RFP. Each is described and analyzed separately within the body of this report.

• Project Influence
• Financing
• Market Conditions (Time)
• Property Size
• Location
• Land Use Regulations
• Road Frontage/Access
• Mineral Rights
• Severable Use Rights and Transferable Development Rights

Our investigation and analysis during the performance of this market study uncovered no additional variables that appear to have a significant impact on land values within the East Everglades Expansion Area.

The report is organized to directly address the variables the Department of Justice requested be included in the study. The study took place between June 1, 1997, and December 31, 1998, as revised, and involved four phases: (1) database creation, (2) property inspections, (3) research and analysis, and (4) report preparation. The first phase involved establishing the descriptive database. The sales data was mapped and geocoded on geological survey quadrant maps using longitudes and latitudes up to four significant digits. The properties were inspected either on foot, by swamp buggy, by road vehicle, or by helicopter. Where possible and appropriate, the sales were verified in person or by telephone.

A data book containing a significant sampling of the sales pertinent to this market is included under separate cover and is considered an integral part of this study. Each of the sales was input into the descriptive database utilizing the input form as illustrated in the Addenda as Exhibit A. A sales data sheet that compiles this information for many of these transactions is included in the data book and a sample output form is identified as Exhibit B in the Addenda. Also included with this study is a copy on CD of the database in three forms: 1) the main descriptive database in Microsoft Access software; 2) the statistical analysis data set in Microsoft Excel software; and 3) the paired data analysis data set on Microsoft Excel software. All of the sales directly referenced in this report by Pritchett, Ball & Wise (PBW) Number are reproduced in the data book.

Each variable was analyzed. The findings and conclusions that relate to the East Everglades are presented in this narrative report unless otherwise stated. [The findings and conclusions that relate to the Big Cypress National Preserve are contained in a separate report.] The above processes involved a total of 12 professionals: four MAI designated appraisers, one database management consultant, two statistical market analysis consultants, three senior consultants/appraisers, and two staff administrators.

II. OVERVIEW
DESCRIPTION OF SALES ANALYZED FOR BOTH EAST EVERGLADES AND BIG CYPRESS

The database was developed utilizing sales from appraisals of Florida's freshwater wetlands provided by the Department of Justice, the National Park Service, the Corps of Engineers, and the South Florida Water Management District. The appraisals were prepared by the agency's staff appraisers and/or independent fee appraisers. There were a total of 1,129 sales originally considered. After deletions and adjustments to the database (as described in more detail later in this report; see - "Description of Sales Analyzed"), the descriptive database contains a universe of 785 sales, dating from 1972 to 1997. Many of the sale write-ups were of the same transaction as investigated and reported by different appraisers. The publicly recorded factual information was generally the same for these "duplicated" sales, but the comments resulting from the various appraiser's interviews with grantors, grantees, brokers, and other knowledgeable market participants may have differed. We consolidated the various observations and opinions in order to report as full a picture as possible about the conditions of the sale and the motivations of the buyer and seller.

We eliminated duplicates, any sales that included significant structures or improvements, and sales that involved any government agency or entity with the power of eminent domain. This reduced the total database to 785 comparable sales. The technical framework of the database is described in the following section of this report.

To accommodate separate studies and reports for Big Cypress and East Everglades, we divided this data into two databases. The following table provides an overview of the data after this division, focusing on the number of sales transactions in each Expansion Area in their various size ranges. As shown, there are 393 comparable sales within the East Everglades descriptive database and 392 in the Big Cypress descriptive database.

TABLE 1, SALES BY SIZE
Size (Acres)	Big Cypress Sales	East Everglades Sales	Total Sales
0 - 2.5	89	97	186
2.51 - 5.0	140	81	221
5.01 - 10.0	67	62	129
10.01 - 20.0	20	41	61
20.01 - 30.0	3	17	20
30.01 - 40.0	17	28	45
40.01 - 100.0	10	37	47
100.01 - 200.0	21	14	35
200.01 - 300.0	4	3	7
300.01 - 400.0	6	5	11
400.01 - 1,000.0	5	5	10
1,000+	10	3	13
Total	392	393	785
Compiled by Pritchett, Ball & Wise, Inc.

Of the 393 sales in the East Everglades database, 178 are sales of properties consisting of five acres or less. This represents 45% of this database. For Big Cypress, 229 (58%) of the 392 sales were of properties of five acres or less. Graphs which illustrate the data within each database are presented below.

The descriptive database includes a number of transactions to and from Michael Jones for the transfer of severable use rights (SURs), a group of sales that were purchased by an investor (7^th Cavalry Corporation) specializing in real estate tax auction sales, and other non market value transactions, such as sales between related parties. The East Everglades database sales that were eliminated from our analyses, and the reason for elimination, are identified in the following table. The sales remain in the descriptive database for the reference of reviewers and appraisers, but should not be relied upon for comparable sales analysis purposes.

TABLE 2, SALES ELIMINATED FROM STATISTICAL AND PAIRED DATA ANALYSIS DATABASES
PBW No.		Reason for Elimination
1052		Auction of bankrupt corporation
2004		Tax sale
2007		Consolidation among partners
2008		Tax sale
2009		Purchased for SURs
2015		Liquidation
2017		Quit Claim Deed
2075		Related parties
7023		Not an arms length transaction
12007		West Palm County
12152		Broward County
12188		Broward County - commercial gas station
12256		No cash; swap for land in Mexico
12271		Trade for boat parts, not cash
12287		Donation to Micosukee Indian tribe
12306		Donation to Miccosukee Indian tribe-Commercial Land Use
12312		Only commercial property in data / not representative of study
12320		Contract 20 years ago
12325		Contract date unknown
12340		Appraiser's comments said not arm's length
12342		Contract in 1976
12347		Property bartered for groceries
12401		Payment of debt, only asset, debtor left U.S.
12402		Payment of debt, only asset
12404		Distressed sale to son-in-law
12410		Contract for deed
12433		Sale for/of SURs
12436		Purchased for SURs
12438		Sale for/of SURs
12446		Quit claim Deed
12452		Sale for/of SURs
12462		Sale for/of SURs
12466		Tax sale
12467		Sale for/of SURs
12468		Sale for/of SURs
12470		Sale for/of SURs
12472		Sale for/or SURs
12473		Sale for/or SURs
12484		Purchased for house, illegal fill at time of sale
12568		Monroe County
20012		Distressed Sale
20019		Not arms length transaction
20061		Sale for/of SURs
20063		Not arms length transaction
20082		Sale for/of SURs
20083		Sale for/of SURs
20085		Not arms length transaction
20090		Sale for/of SURs
20106		Quit claim deed
20110		Sale for/of SURs
20111	Sale for/of SURs
Compiled by Pritchett, Ball and Wise, Inc.

 

 

TABLE 2, CONTINUED. SALES ELIMINATED FROM THE ANALYTICAL DATABASE
PBW No.	Reason for Elimination
20112	Distressed Sale
20114	Not arms length transaction
20116	Incomplete sales information (sale date unknown)
20138	Not arms length transaction
20139	Not arms length transactions
20140	Sale for/of SURs
Total	57
Compiled by Pritchett, Ball & Wise, Inc.

III. DEVELOPMENT OF THE DATABASES

Descriptive Database

The sales were input into a Microsoft Access database format which consisted of one main sales table, seven subordinate sales information tables, and 11 reference tables. Data on each of the sales was coded within five major headings: location, physical data, sales data, verification, and comments. An attempt was made to reverify the transactions, but given the age of the comparables, some sales could not be reverified with any party to the transaction. All sales provided to us by other sources and relied upon in this study were previously verified by qualified appraisers.

Design

The database was modeled using standard Entity Relationship Diagramming (ERD) techniques to logically model the real estate data entities, attributes, and relationships. The logical data model was then translated into a physical database design by defining table and columns for the logical entities and attributes. Each entity natural key attribute was translated to a table primary key column for unique identification of each table row (record). Relationships were defined between the subordinate sales tables and the main sales table. The corresponding foreign keys were defined between the tables. Various index tables were defined to improve the performance of certain database queries.

The sales table has a "one-to-many" (1-M) relationship with each of the seven subordinate sales tables. In other words, for each main sales record there may be one or more records in each of the subordinate tables. For example, a sale may have one or more legal descriptions depending on how many parcels of land make up the sale.

The purpose of the reference tables is to store the allowable set of values for a field (column) in one of the sales tables. The sales and reference tables are identified in the following chart.

TABLE 3, DESCRIPTION OF TABLES
Table Name	Description
Areas	Options (values): Big Cypress and East Everglades
Buyer_Motivation	Buyer Motivation Land Use
Counties	All Counties In and Near the Project
Deed_Type	Options (values): Quit Claim, Special Warranty, and Warranty
In _Big_Cypress_Exp	Township, Range, and Section in Big Cypress Expansion Area
In_Corps_Project	Township, Range, and Section in Corps Project
In-East_Ever_Exp	Township, Range, and Section in East Everglades Expansion
In_Near_Municipalities	Township, Range, and Section in or near Municipalities
Interest_Conveyed	Options (values): Fee Simple, Leasehold, and Leased Fee
Municipalities	All Municipalities In and Near the Project
Near_Big_Cypress_Exp	Township, Range, and Section near Big Cypress Expansion
Near_Corps_Project	Township, Range, and Section near Corps Project
Near_East_Ever_Exp	Township, Range, and Section near East Everglades Expansion
Near_Howard_Br_TRS	Township, Range, and Section near Howard St. Bridge
Near_Richmond_Br_TRS	Township, Range, and Section near Richmond St. Bridge
Near_SW_288th_Br_TRS	Township, Range, and Section near SW 288th St. Bridge
Near_Tamiami_Br_TRS	Township, Range, and Section near Tamiami Trail Bridge
Part_Type	Partition Type (Ex. E1/4, SW1/4)
Regulations	All Possible Applicable Regulations
States	U.S. States, Canadian Provinces, and Mexican States
Use_Type	Land Use Types
Water_Type	Types of Water Frontage
Compiled by Pritchett, Ball & Wise, Inc.

A detailed listing of the subordinate tables is located in the Addenda, labeled as Exhibit K.

A data file over 41,000 ownership records in the Dade County data file used by the county tax assessors was received from Dade County on CD-ROM. The data file was converted into a Microsoft Access database. The data was used to verify and complete the sales data, particularly the Grantee information, on the Sales database for the East Everglades.

Sales were matched to the Dade County data file by two methods: 1) folio number and 2) owner or Grantee name. Database queries were written to do the matching, matches were hand-checked, and data was taken from the Dade County data file to enter into the Sales database. After data entry for the main Sales database was completed, the data for the East Everglades and Big Cypress areas was converted into separate outputs for statistical analysis by the SPSS (Statistical Package for the Social Sciences) software. The structure of the selected sales consisted of a set of variables that were defined to SPSS standards for the analysis process (the statistical study).

Each variable in the statistical study was derived from the main Sales database. Other variables were defined between the descriptive database and the selected statistical data to associate two or more variables. For example, variables were defined for property size and accessibility.

Another set of variables was defined to associate size and accessibility, such as 1 to 3 acre sales on all weather roads or 1 to 3 acre sales on dirt/grass roads. These associative or interaction variables were used by SPSS to determine any significant statistical relationships between and among the variables.

There are 336 sales in the data sets used for statistical or paired sales analysis. Of these, 229 are within the Overlay Ordinance (any Management Area). Only three management areas, MA 1 (34 sales), MA 2A (98 sales), and MA 3B (51 sales) proved to have a direct influence on the price of land. Tracts in MA 2A sold for significantly less per acre, all other variables accounted for. Tracts in MA 1 and MA 3B generally sold for more per acre, but the differences were not always statistically significant at the p = 0.05 level of significance. The other management areas (MA 2B, MA 3A and MA 3C) were not statistically significant variables. However, we concluded that in general the market for lands controlled by the Dade County Overlay Ordinance is different than the market for similar land wherein land use is not as closely controlled.

One hundred and seventy-seven (177) sales are in the East Everglades Expansion Area, 58 are "near" (within one mile of) the Expansion Area, and 101 are "not near" (more than one mile from the Expansion Area). These variables were used as a means of estimating potential project influence, and they also help in understanding the impact of the regulatory environment both within and outside the areas governed by the Dade County overlay ordinance. Our analysis shows that the announcement of the federal projects (the expansion of East Everglades National Park and /or the Corps of Engineers project) did not have an influence on the price paid for land.

Eighty-four (84) of the sales were of property used for some type of agriculture at the time the tract was sold or, if agricultural use at the time of the sale could not be determined, at the time we inspected the sale. The rest of the tracts were best classified as a limited recreational land use. The agricultural sales sold for a significantly higher price per acre than the other sales. Consequently, we generally partitioned the sales by land use and made separate comparisons for the agricultural and non-agricultural sales.

Ninety (90) of the sales indicated purchase money notes or other owner financing. The balance of the sales were for cash or third party financing. Very few of the comments indicated that any third party lenders were involved in smaller acreage transactions. The evidence seems to show that owner financing generally increases the price paid for wetlands, even if the terms of the owner financing are similar to terms offered by third-party lenders.

Access to the property is one of the more important variables affecting value. 81 sales have direct access to a paved road and 23 sales on a gravel/limestone road. Both of these types of access affect the price paid for the land. Seventy-seven (77) of the sales indicated a dirt road or grass road access, 76 via either airboat or buggy, 10 by airboat only, 5 by buggy only and 64 with no vehicular access, in which case access is via either helicopter or walking. There is some evidence that direct access by a dirt/grass road is superior to airboat and buggy access, but no evidence that airboat and/or swamp buggy access is superior to no known access.

A total of one hundred and seventy (170) of the sales were within a mile of a paved road, but, by itself, this variable was not significant. However, for parcels without an all weather direct access to the property, proximity to a paved road appears to be a positive influence on price.

Fifteen (15) of the sales were within a mile of one of the four bridges that provide access from the east to the west of the L31-N Canal. This variable made a significant difference in the price paid for land. In some cases, being within a mile of both a major road and a major bridge affected the value of the land.

IV. THE ANALYTICAL MODELS

The analytical data is made up of sales of vacant tracts of land in the freshwater wetlands of south Florida. Much of this land is under water for between three and nine months of the year. Many of the tracts are inaccessible and have very little identifiable functional utility. It is difficult to ascribe rational, economically based motives to the buy-sell decisions. Consequently, there is an unusual degree of variability in the data. Two tracts that appear to have virtually the same objective criteria (date of sale, size, location, physical characteristics, etc.) will often have sold at very different prices. The problem for the analyst is to be able to distinguish between the objective criteria that actually influence the price paid for these freshwater wetlands from the "noise" that comes from a relatively inefficient, unusual real estate market wherein all of the buyers and sellers are not equally well informed or well advised.

One of the essential characteristics of real estate is that it is not fungible. Each parcel is uniquely fixed in space. At the same time we still believe that the economic principal of substitution applies. That is to say that the market value of a property can be determined by comparing the sales of other parcels with similar physical characteristics and similar functional utility. This is the concept embodied in the sales comparison or "market" approach to the appraisal of real estate, and the sales comparison approach is the primary approach applied to the appraisal of vacant parcels. Basic appraisal practice requires that the appraiser select at least three "market value" sales of reasonably comparable properties, and that he or she adjust the comparable sales to make them more like the subject. The adjustments are to be based on indications from the market about the direction and magnitude of value differences associated with observable differences wherein the comparable property is considered to be superior or inferior to the subject property. [For a more detailed discussion of the sales comparison approach, see generally THE APPRAISAL OF REAL ESTATE, ELEVENTH EDITION, The Appraisal Institute, Chicago, IL., 1996, Chapters 18 and 19, pp. 397-446.]

The process of selecting and adjusting comparable sales is an exercise in judgment for the appraisers, but we believe it mirrors the calculus of the market of willing buyers and willing sellers, each equally well informed or well advised. It is important to consider at least three sales because three sales are the fewest number that will indicate the degree of central tendency and dispersion in the market. If all three sales are reasonably similar (i.e. require few adjustments) and if all three indicate about the same price (i.e. indicate a strong central tendency and little dispersion), then the buyer, seller and appraiser generally agree that the market value of the subject is about the value indicated by the adjusted value of the three comparable sales. On the other hand, if the adjusted value of the three sales shows a relatively wide dispersion, the buyer, seller and appraiser are generally dissatisfied and wish to seek out more sales to better indicate the magnitude and direction of adjustments and to better verify the central tendency reflected by the judgment of the marketplace.

Appraisal judgments are based on the appraiser's broad knowledge of the market and are reinforced and refreshed by interviews with market participants as a background to his or her selection and examination of comparable sales. In most cases, when the appraiser is fortunate enough to be working with property in a relatively rational, economically based market, differences in price can be rationally explained by observable differences that objectively reflect differences in the functional utility of the different parcels. Generally speaking, the important differences in functional utility can be uncovered by a careful matched pairs examination of fewer than ten comparable sales. The adjustments can be rationally defended; thereby reducing the dispersion in the data, and the indication of central tendency in the data is clearly reflected by the appraiser's judgment about the market value of the subject property.

Unfortunately, the sales of south Florida's freshwater wetlands have very few of the characteristics of a rational marketplace. Even though there are several hundred sales in the database, there is a wide dispersion of prices within the data and the differences in functional utility between parcels are as extreme.

An analyst faced with an array of data seeks to find information from the data by using measures of central tendency and dispersion. Most of the time, the analyst makes judgments about the information contained in the data based on the indications of central tendency. The three most common descriptions of central tendency are the mean (mathematical average), the median (mid-point of the array), and the mode (most frequently occurring value in the array).

The most common descriptions of the measure of dispersion are the range or extremes of the array and the standard deviation of the data points relative to the mean. The smaller the range and the smaller the standard deviation, the greater the degree of confidence that the analyst can place in judgments based on the central tendency.

The mean and standard deviation are also the most useful descriptive statistics because they are mathematical constructs, which means that they can be measured and compared mathematically. This ability is particularly helpful with data, such as that generated by the subject, where "answers" are not readily apparent.

TABLE 4, EXAMPLE OF MEAN AND STANDARD DEVIATION FROM TWO DATA ARRAYS
		MARKET A;	MARKET B
SALE NO.		$/ACRE	$/ACRE
1		$7,250	$5,250
2		$6,450	$4,950
3		$6,150	$5,150
4		$5,350	$7,250
5		$5,500	$4,750
6		$5,000	$5,000
7		$4,200	$5,050
8		$4,200	$2,500
9		$3,400	$5,150
10		$2,500	$4,950
MEAN		$5,000	$5,000
STANDARD DEVIATION		$1,377	$1,072

For example, if we partition or split the array of data into two sets based on some observed difference, such as date-of-sale in the hypothetical example above, we may find that both sets have the same average price per acre (a mean of $5,000/acre), but that the standard deviation of one set (Market B) is substantially smaller than the standard deviation of the other set (Market A). A comparison of the two average values alone may lead to an erroneous judgment that the observed difference in market areas doesn't matter, whereas a comparison of the two standard deviations shows that difference in market area probably does affect the price per acre.

A mathematically based statistical analysis could make use of the difference in the standard deviation of the two arrays to assign a probability that any observed relationship (average value given market area in this example) was just due to chance for Market A, but that it was statistically significant (probably not due to chance given a stated confidence interval) for Market B. If all we know about the two groups of sales is the sales price/acre, the only statistics we can use to "best" describe the data are statistics about central tendency like the average, the median and the mode(s) and statistics about dispersion like the standard deviation and the range.

The analysis of data from the south Florida wetlands was performed utilizing three basic methods: statistical analysis, paired data analysis, and qualitative analysis. Each method is described in the following paragraphs.

Statistical Analysis

All analytical methods use statistics to a greater or lesser extent. If we compare matched pairs of data wherein we have more than one pair, we will generally base our judgment on the central tendency and dispersion shown by the multiple matched pairs (i.e. "Most of the matched pairs indicate an adjustment of $X"). If we base our judgment on interviews with knowledgeable market participants, we generally report that "a majority of" the market players said thus and so. In the statistical analysis component of the market study we are much more rigorous and formal in the foundation of our judgment, because we actually specify our confidence interval and make our judgment about the variables that affect value by following a body of mathematically based statistical rules and techniques.

The size and complexity of the descriptive database strongly suggested the use of statistical analysis. We employed a stepwise, multi-linear regression analysis to uncover trends and relationships subtly embedded in the data. This methodology has a strong history in the study of real estate and, when used as an adjunct to rather than as a substitute for expert judgment, can be an excellent tool for a market analysis.

The object of regression analysis is to examine the relationships between a dependent variable, such as sale price or price-per-acre and a set of independent variables with which sale price or price-per-acre may be associated. The independent variables of interest are those factors that appear to "cause" or determine the differences in the prices paid for land in the Florida freshwater wetlands. Each potential independent variable was studied in turn by modeling its relationship to the dependent variable (price or price-per-acre). Most of these factors are qualitative rather than quantitative in nature. For example, "location" is a variable that is not naturally numeric but must be converted into a numeric form before the regression analysis could be conducted. The same point applies to many of the factors under study. The qualitative information contained in the database therefore was converted into either a set of binary variables (0,1) or numeric variables (number of miles from a paved road or the number of months from the date of the first sale).

Once all information has been quantified, regression analysis calculates equations that best account for the variability in the dependent variable. The regression equations are attempts at explaining why price per acre varied based on changes in the studied factors ( e.g., location, size, land use controls, etc.). The algorithm used to calculate the regression equations is the least-squares algorithm, and the statistical package employed was SPSS, one of the statistical packages commonly employed by social scientists.

The following hypothetical example shows the same two arrays of data presented above with the addition of data on the size (in acres) of the sales.

TABLE 5 EXAMPLE OF MEAN AND STANDARD DEVIATION FROM TWO DATA ARRAYS
		MARKET A	MARKET B
SALE NO.	SIZE	$/ACRE	$/ACRE
1	1.75	$7,250	$5,250
2	2.25	$6,450	$4,950
3	2.5	$6,150	$5,150
4	3.5	$5,350	$7,250
5	3.75	$5,500	$4,750
6	5.25	$5,000	$5,000
7	6.5	$4,200	$5,050
8	6.5	$4,200	$2,500
9	8	$3,400	$5,150
10	10	$2,500	$4,950
MEAN	5	$5,000	$5,000
STANDARD DEVIATION	2.58	$1,377	$1,072

In Figure 3 showing the relationship between size and price/acre for Market A (Table 5, above), the scatter diagram shows that most of the data lies on a downward sloping line. This indicates that larger parcels sell for a lower price-per-acre than smaller parcels. The straight line drawn through the data points is called the "best fit" line or the "least squares" line. It represents the point at which the sum of the squared difference (deviation) between the line and each observed data point (sale) is at a minimum. The formula [y=-471.52x + 7593.3] is the formula for the line {Y = A + B(x)} wherein B(x) [471.52 (x)] describes the slope of the line and A [7593.3] is the constant or point at which the line intercepts the Y axis.

R² is a statistic that describes the extent to which variations in "size", which is the independent variable, explain the variation in "$/acre", which is the dependent variable. In this example 97% of the variation in $/acre is "caused" or explained by differences in the size of the parcel.

The following Figure 4 shows the relationship between size and price-per-acre for the array of data labeled Market B.

In this case the scatter diagram of sales does not show the clear pattern of a strong relationship between size and $/acre indicated by Market A. All but two of the sales fall roughly along a line representing about $5,000/acre, regardless of the difference in the size of the parcel. We calculated a "best fit" line or regression line through the data points [y=-126.67x + 5696.7], but the R² calculation is only 0.1152, which is to say that only 11.5% of the variation in the $/acre is associated with or explained by differences in the size of the parcels.

If we were to add a test of the statistical significance at the p = 0.05 confidence interval to the regression model for Market B, we would be able to calculate a "p" value for the regression of p= 0.72. The value of "p" for any regression is a function of the amount of explained and unexplained dispersion in the data given the size of the sample and the number of variables being considered. With a "p" value that is as high as 0.72, we would reject the observation that difference in size was correlated to $/acre, because we would believe that the perceived weak relationship was due only to chance (p = 0.72 which is > p = 0.05).

To sum up what we hope to show using this hypothetical example, in both cases we started with an array of ten sales. These arrays each have a measure of central tendency or average value, or mean of $5,000/acre. They both had a measure of dispersion or range of between $2,500/acre and $7,200/acre. In the case of Market A, our understanding of the difference in $/acre among the ten sales is substantially (97%) explained by differences in the size of the parcels. In the case of Market B, very little of the difference is explained by differences in the size of the parcels.

An analyst given only the data shown in the hypothetical example above would be able to conclude that best indication of the market value of the Market B sales was $5,000/acre, and that differences in the size of the tract (within a range of from 1.75 acres to 10 acres) do not affect the $/acre. However, for the Market B sales, although the average price was also $5,000/acre, the actual value of any single tract was substantially affected by differences in the size of the tract. As size increased (or decreased) from the average of 5 acres, $/acre decreased for the larger sales and increased for the smaller sales by about $470/acre.

Correlation is not the same as causality. A comparison of the two samples may indicate to the analyst that there is a difference between Market A and Market B. However, there is nothing in the data that explains why a difference in size of parcel affects value in Market A whereas it does not affect the value in Market B. Explanations of causality are matters of judgment, not statistics.

This is an example of simple linear regression. Multiple linear regression examines a range of potential explanatory variables, each of which is theoretically independent of each other explanatory variable. The explanatory power of each statistically independent variable is added to the explanatory power of all other statistically significant independent variables to calculate the multiple R² for the equation. Whereas the equation for a simple linear regression was Y = A [the intercept or calculated constant] + B(X) [being the slope of the line], the multi-linear regression model is A + B₁(X₁) + B₂(X₂) +B_n(X_n) where A represents the calculated constant or average value as-if none of the X variables affected the dependent (Y) variable, and the B subscripts represent the slopes of the "best fit" line for each of the "n" independent (X) variables to be considered.

The "calculated constant" for a multi-linear regression is similar to the intercept for a simple linear regression model. It represents the "A" in the regression model Y = A + B₁(X₁) + B₂(X₂) +B_n(X_n). If one could visualize a multi-linear regression in two dimensional space, it represents the point at which the "best fit" line crosses the Y access. If none of the n number of "X" variables affected the Y variable, that is to say that there are no significant independent variables, the calculated constant represents the "average" or "typical" value of the dependent variable (say $/acre). In a regression model wherein there are statistically significant X variables, the calculated constant represents the average value before considering the influence of the statistically significant independent variables.

The calculated equations represent regression's "best guess" at the underlying relationship between the dependent variable and the independent factors. Regression has a way of expressing its confidence in these "best guesses". The p-value is a ratio ranging from 1 to 0 that expresses random rather than true relationships (noise rather than correlation).

A high p-value indicates that there is little confidence in the estimated equations, that they are probably just random events and that they do not represent any strong relationship between the independent or "explanatory" variable and the dependent variable. This is what is meant by the term "statistical significance". A p-value of .05 or less would indicate that there is a 95% or greater probability that the observed relationship impacts value (or more properly, that there is only a 5% or less probability that the observed relationship is due to random events).

In the sections which follow, the results of the regression analysis and their statistical significance will be discussed. Rather than present equations, the discussions will focus on the practical interpretations of the regression results. At its best, the statistical model shows correlation, not causality. That is to say that the statistics can document that a relationship exists between the variables. However, the judgment about the causative nature of the relationship between (or among) the variables reflects the judgment of the analyst.

The stepwise function multi-linear regression model was used to consider the potential impact of the independent variables (such as type of road providing access to a parcel) and interactions between variables (such as type of access and proximity to a major road). "Stepwise" means that each potential independent variable was tested in every possible combination with all other potential independent variables to test for problems of co-linearity (two possible independent variables related to each other). "Multi-linear" means considering multiple independent variables each adding their impact to "cause" changes in the dependent variable. At the 0.05 level of significance, eleven (11) of these variables "qualified" as being independent variables that "caused" changes in the price of land. The Multiple R (best fit statistic, which is sometimes called "Pearson's r" or the correlation coefficient) from our "best" model is 0.80682, and the R² statistic is 0.65. Whereas the Multiple R is a statistic about other statistics and does not translate into a "real world" useful meaning, R² can be understood as the percentage of the change in the dependent variable (price) explained by changes in the independent variables. In the East Everglades analysis, 65% of the difference in land prices is explained by the statistically significant variables.

We created a statistical data set from the 336 sales that we believe represent market transactions that may indicate the variables affecting the value of the East Everglades Expansion Area. We identified 177 potential independent variables (including interaction variables) for each of the sales, plus several potential dependent variables, primarily price per acre ($/acre) and total transaction price (price). Each sale is identified by a unique number that we assigned (PBW No.) and another sequential number assigned by SPSS. This 336 row by 181 column data set is reproduced as an Excel spreadsheet called the East Everglades Statistical Data Set (East_e~1.xls). A copy of the statistical data set is reproduced on the CD that accompanies this report.

A simplified version that lists only the 32 basic potential independent variables (not counting interaction variables) was developed to permit matched pairs comparisons. This 336 row by 34 column data set is reproduced on the CD as an Excel spreadsheet called the East Everglades Matched Pairs Data Set (Eerevise.xls). Using the Excel software's "filter" command it is possible to recreate any interaction variable (i.e. less than 3 acre sale after the East Everglades Expansion announcement but before the Corps of Engineers announcement located on a paved road in Homestead City near a major bridge).

The following is a list of the statistically significant independent variables in order of importance.

TABLE 6, SIGNIFICANT VARIABLES
Variable		Probability Outcome Due To Chance
SIGNIFICANT VARIABLES FOR >3 ACRE TRACTS; $/ACRE AS DEPENDENT VARIABLE
Agricultural Use		p = 0.0000
6-15 Acre Tract Near a Bridge		P = 0.0008
Later Sales (after 5/92 following both project announcements)		P = 0.0003
Management Area 2A		P = 0.0010
6-15 acre Tract with Paved Road Access		p = 0.0008
100 Acre or Larger Tracts		p = 0.0076
15 to 55 Acre Tracts Near a Bridge		p = 0.0000
Not Near Bridge Or Paved Road		P = 0.0397
Size As A Continuous Variable		P = 0.0274
Near (within 1 mile of) a Bridge and Major Road		p = 0.0397
15-55 acre Tract with Grass/Dirt Road Access		p = 0.0237
SIGNIFICANT VARIABLES FOR <3 ACRE SALES; TOTAL SALES PRICE AS DEPENDENT VARIABLE
Near A Bridge	P = 0.0000
Near (Within One Mile Of) The Expansion Area	p = 0.0000
Compiled by Pritchett, Ball & Wise, Inc.

One of our most important findings is that total transaction price (price) rather than price per acre ($/acre) is the most appropriate dependent variable for the less than three acre sales (<3 acre); whereas price/acre ($/acre) is the most appropriate dependent variable for the three acre or larger (>3 ac.) sales.

Florida uses the government survey method for legal descriptions of parcels, rather than the metes and bounds system. A natural consequence of this system is that parcel sizes tend to be clustered at the points best described by the survey system, that is at 1.25 acres, 2.5 acres, 5 acres, 10 acres, 20 acres, 40 acres, etc., rather than evenly distributed along a continuum of size. As one can see from an examination of the scatter drawing below, the 71 sales of parcels that are less than three acres in size are primarily located at 1.25 acres, and at 2.5 acres.

When we use $/acre as the unit of comparison (dependent variable), the average $/acre is approximately $5,000/acre, regardless of the size of the parcel. The "best fit" line that can be drawn through the data is almost a level line (no slope) and the calculated R² is very small (0.0002), showing that differences in size among these <3 acre sales is not useful as an explanatory variable.

The data presented in Figure 6, below, shows the same <3 acre sales using total sales price as the dependent variable. In this case the "best fit" line slopes upward slightly, showing that buyers paid slightly more for 2.5 acre tracts than they did for 1.25 acre tracts, but only slightly more. The calculated R² for the data using total sales price is 0.08, showing that as the size of the parcel increased from 1.25 acres to 2.5 acres, the price increased about 8% from the average price of about $5,000 for a 1.25 acre tract.

If $/acre was the appropriate unit of comparison for these smaller sales, a 2.5 acre tract would sell for twice as much as a 1.25 acre tract. In that case we would expect to see an upward sloping line at a 45 degree angle and a close relationship between the observed sales and the line. Figure 6 shows that the slope of the line is only at about a 15 degree angle and that the data is not closely grouped around the line.

We further tested the idea that total sale price was a more meaningful unit of comparison for the <3 acre sales, whereas $/acre was a more meaningful unit of comparison for the >3 acre sales, by running full multi-linear regression models on all of the data, on the <3 acre sales and on the >3 acre sales. When we used $/acre as the dependent variable for all sales the calculated R² is 0.55. When we split the data into two data sets and calculated the R² for the smaller sales, R² was reduced to 0.52, which is a very minor change (5% inferior). However, the R² for the larger sales was increased to 0.65, which is a substantial improvement in the explanatory power of the regression model (18% superior).

Given the problems of identifying functional utility in these wetlands in the first place, it is not surprising that in many cases a "useable" parcel may have to encompass 2.5 acres, rather than 1.25 acres, in order to have a tract that contains a hammock or other desirable feature. It appears that these buyers may be willing to pay a little more for a 2.5 acre tract than for a 1.25 acre tract, but not much more. Once we discovered that the appropriate dependent variable was different for the <3 acre sales than it was for the larger sales, we split the statistical data set in order to appropriately study the smaller sales as distinct from the larger sales.

We used the regression model to forecast the prices for each of the sales, and then examined the "residuals" (the differences between the prices that the regression model calculated and the actual sales prices) to see if there are any other patterns than might indicate other explanatory variables. No additional variables were significant from this study of the residuals.

In summary, we believe that the regression model is a useful tool to assist in an understanding of the variables that show evidence of affecting the price of the freshwater wetlands in south Florida, and in promoting consistency in appraisal judgments by focusing attention on the variables that have the greatest influence. With our study we have identified 11 important variables, and eliminated 166 potential variables as items that do not warrant careful attention. With it we can explain 65% of the variation in the observed price of parcels that sold between 1972 and 1997. At the same time, we believe that traditional paired data comparison is a useful and persuasive method to value any single parcel. However, the appraiser who bases his or her conclusion on paired sales can not ignore the variables identified above and still claim to be using sales that are comparables.

Paired Data Analysis

Introduction

In an ideal situation, paired data analysis is a procedure in which two properties are compared for the purpose of determining the effect on value "caused" by an observed difference between the properties. To start the analysis, two comparable properties or sets of comparables that are identical in all characteristics except one are matched as pairs. With these two sets being identical except for one characteristic, the difference in the sales price per acre is attributed to the single difference.

The Appraisal Institute defines Paired Data Analysis in The Dictionary of Real Estate Appraisal on Page 258 in the following way:

A quantitative technique used to identify and measure adjustments to the sale prices or rents of comparable properties; to apply this technique, sales or rental data on nearly identical properties are analyzed to isolate a single characteristic's effect on value or rent.

This procedure was formally known as a matched pair analysis.

In a real world situation, the ideal setting described above is not usually attainable. Paired data analysis must be used with caution when the analysis must be performed in less than ideal conditions. It is usually not possible to find two sets of comparable properties that are identical in all but one characteristic. At best, the analyst will be able to find:

1. two comparable data sets that are similar with regard to several major characteristics but are different with regard to many minor characteristics. In this situation, a price difference can only be attributed to the array of minor characteristics that differ between the two comparable data sets.
2. two comparable data sets that are similar with regard to several major characteristics but are different with regard to one major characteristic plus many minor characteristics. In this situation, a price difference can be attributed to the difference in the major characteristic as well as the array of minor characteristics that differ between the two comparable data sets.

The paired data analysis performed in this study was constrained by these two limitations. It was also constrained by the quantity and the quality of the sales information available to the analysts. Even though we have several hundred sales in our statistical data set, we quickly run out of sales when we try matched pair analysis on multiple variables. In some cases we were able to develop matched pairs using three variables, but usually we ran out of data on any match with more than three variables.

Anyone who makes judgments about the market value of any parcel in the freshwater wetlands of south Florida must recognize that there is a great deal of unexplained variation in the sales data. It would be possible to select a "high" sale (or a small group of "high" sales to be compared to a "low" sale (or small group of "low" sales) on some supposed variable that explained market behavior in a matched pair and come to an erroneous conclusion about a perceived difference, due only to the amount of variability in each group of sales.

Identification of the Major Characteristics for the Paired Data Analysis

We established a list of the variables expected to be major influences in the analysis. This list was based on the collective experience and knowledge of the principal investigators with regard to land sales. This list was then compared to the list of characteristics as outlined in the RFP. All of the essential elements of the RFP were captured in the study.

The variables that proved to be important are:

1. Land Use-Agricultural or Recreational
2. Management Area (Land Use Regulation in Dade County)
3. Date (Sales After Both Announcements)
4. Access Characteristic (Type of Direct Access to the Property)
5. Property Size (Both as a continuous Variable and by Discrete Size Categories)
6. Location Characteristic (within one mile of a major bridge and/or a paved road)

The Data Set

The data set for the paired data analysis is the same as for the statistical analysis, with the exception that we eliminated the interaction variables and the variables for which we had no sales. This made a manageable data set of 32 variables for each sale. Any of the interaction matrices can be recreated using the "filter" function in the Excel spreadsheet program. Several of the property characteristics appeared in too few cases to be useful. This was the case with the following variables: (1) Management Area 2B, which contained 1 data point, (2) "near" (within one mile) of a municipality, which contained zero data points; and (3) West Palm, Broward and Monroe Counties which contained no (0) data points, respectively.

For example, we examined the importance of the Dade County Management Area Overlay Ordinance. Table 7 shows a preliminary partition of the data into sales of parcels controlled by the Overlay Ordinance and sales outside any of the management areas.

The substantial land use restrictions imposed by the Dade County Management Area Overlay Ordinance have had a pronounced impact on the value of the land in the freshwater wetlands. As we explain more fully in a later section of this report, Dade County passed a Zoning Overlay Ordinance in 1982 substantially limiting development in most of the freshwater wetlands to the west of the L31-N canal. There are six Management Areas established by this ordinance: MA 1, MA 2 A and 2 B, MA 3 A, 3 B, and 3 C.

We examined approximately 229 sales from within Management Areas, as compared with 107 sales not in a Management Area. We further split the data into the 71 small sales wherein total sales price is the appropriate unit of comparison and the 265 larger than 3 acre sales wherein $/acre is the appropriate unit of comparison. The study shows clearly that the market for land not governed by the Overlay Ordinance is different, and that sales not in a Management Area (MA) should not be used as comparables to indicate the value of land within a MA. This conclusion is hardly surprising. The Overlay Ordinance has had the practical result of restricting the opportunity to build a residence to properties consisting of at least 40 contiguous acres.

Use of Paired Data

We believe that, given the variability within the data, it is not reasonable to undertake a paired data analysis unless there are at least three sales that represent the qualities that one is attempting to hold constant. Often, once the data is sorted on as few as three variables, there are no longer at least three sales that represent the match criteria. In most cases we show the data as a table, reporting the number of observations in a cell, the average and the standard deviation. In some cases we show the results as a scatter diagram wherein we can show the correlation or lack of correlation. In a few cases we show the actual data in the body of the report. The actual paired data upon which the table or chart is based is provided on the accompanying CD.

Qualitative Analysis

The qualitative analysis represents anecdotal evidence from our discussions with the sellers and buyers of properties, real estate brokers, federal, state, and local government officials, and other knowledgeable market participants. It also includes our analysis of published information obtained from numerous public and private sources. Anecdotal evidence from knowledgeable market participants and observers is an independent method of seeking to learn about the variables that may impact value.

V. PHYSICAL DESCRIPTION OF THE EVERGLADES

The freshwater wetlands that make up the Everglades stretch over 4,000 square miles from Lake Okeechobee to Florida Bay. The area has a southern oriented slope of less than three inches (3") per mile. The very slow sheet flow of surface waters, coupled with the periodic wet and dry conditions, create a unique ecology. Maps illustrating the East Everglades along with the Everglades National Park boundaries and Expansion Areas are in the Addenda as Exhibits C and E.

The area of primary concern in this report is the approximately 109,000 acres to the north of the Everglades National Park, to the east of Big Cypress National preserve, to the south of US Route 41 (Tamiami Trail) and to the west of the L31-N Canal. This area is known as the East Everglades Expansion Area and approved for acquisition by Congress in December 1989.

The 1979 edition of the Dade County Comprehensive Development Master Plan (CDMP) contains a detailed description of the area's geography and ecology that is very useful for a basic understanding of the physical and environmental constraints that affect the land use of the freshwater wetlands. The 1997 edition of the CDMP contains detailed information on the land use regulations and other governmental constraints to access, road maintenance and the development potential of the freshwater wetlands areas.

The Natural System

South Florida is a sub-tropical, wet and dry climate. The average annual rainfall is about 60 inches, about seventy percent (70%) of which falls between June and October. During the rainy season, most of the freshwater wetlands flood for between six and nine months. Because the slopes in the area are so gentle, the largest percentage of the floodwaters evaporate, both directly from sunlight and from the transpiration of the plants. Other waters percolate through the limestone substrata to recharge the Biscayne aquifer. Additionally, there is a gentle, shallow sheet flow of the surface water towards the sea through the sloughs, primarily Taylor Slough at the southern tip of Dade County, and through the transverse glades into Biscayne Bay. (CMDP, '79, p. 23 & 38) Exhibit F shows generally the major drainage patterns in south Florida.

Exhibit G shows the primary drainage network affecting the East Everglades. It shows the major canals, levees, impoundment areas and drainage structures constructed over the past 100 years. The primary structures affecting the Everglades National Park (ENP) and the East Everglades Expansion Area are the L-31 N levee and the C-111 canal. Recognizing the influence of both the natural systems and the manmade systems, about 32% of the rainfall is accounted for by surface evaporation, 5% by consumptive water use, 9% by seepage to the ocean, 27% by irrigation, evapotranspiration and private wells, and 27% via the canals. (CDMP, '79, p.35)

The Geology

The geology of Florida is a thick layer of porous limestone sediment. The entire mainland of Dade County is underlain by Miami limestone (a mix of oolites and bryozoan deposits characterized by its extreme porous condition and a high degree of water transmissivity). The Everglade Plain, generally west of the L-31 N levee and east of Big Cypress Preserve is very flat, with elevations ranging from a high of between 9' and 10' above mean sea level (MSL) to one foot at Florida Bay. Most of the Everglades National Park is at 2' MSL. The elevation of the coastal ridge ranges from a low of 10' to 15' at the Broward and Dade county border to a high of 23' at Coconut Grove and is broken by shallow valleys called transverse glades. Most of the drainage canals have been constructed in these glades.

The soils in this area are generally classified as sands, rockland, marls, peats, mucks and manmade (resulting from "rockplowing", which is the crushing and mixing of the rockland with the organic soils). In many areas the overburden has been washed away (or never formed) and the surface of the ground is the rough, eroded outcropping of the oolitic or bryozoan limestone. The marl soils were formed from the particulates of calcium carbonate, whereas the mucks and peats were formed of the partially decomposed plant materials. The plant life variations are closely associated with the composition of the underlying soils. The peats and mucks generally support the sawgrass, mangroves and cypress, whereas the bluegreen algae mats are generally associated with the marls.

The limerock itself constitutes the most significant mineral resource in Dade County. It is used as the base material for roads, for landfill, as construction aggregate, and in the manufacture of concrete. The mining costs are minimal, as there is little overburden, and mining is conducted via open pits. The location of the mines is close to the sites where the material is used. The open pits become lakes. There are more than 350 such lakes in Dade County, covering more than 7,000 acres, and about half of the lakes have become the site for residential developments. Many of the existing quarries have permits that will allow them to operate over the next twenty or thirty years. (CDMP, '79, p.33-34) A copy of the area of past and present mining activity and of the deposits of Miami Limestone and of High Hardness Limestone from the CDMP, '97, p. I-74 is reproduced as Exhibit H in the Addenda.

The primary source of the information in the following rock mining paragraphs was provided by the Dade Environmental Management Division (DERM) biologist, Mr. Michael Spenelli. Rock mining is presently not allowed in the East Everglades, either in the Everglades National Park (ENP) or in the Expansion Area. This is generally the area west of L-31 N Levee. The constraints are imposed by the Overlay Ordinance and by the Comprehensive Development Master Plan (CDMP). Only one company, The LaFarge Company in Section 26 & 35, T 54, R 38 (National Park Service Parcel numbers 110-1 and 110-2), about 2 to 3 miles south of SW 8th Street immediately west of L-31, is trying to obtain a permit, and they are running into a great deal of difficulty.

In the areas where rock mining is permitted, the permit process is rather simple. The mining company presents a site plan and a mitigation plan to the county. The major controversies have been over the locations of mines and mitigation plans. Mining is generally permitted, and there are no governmental constraints on the depth of digs, although present technology generally limits depth to 80' to 85'. The State Legislature is considering a "lakeland" district plan to govern rock mining in a more comprehensive manner, based on a more complete understanding of the impact of rock mining on the hydrology and the biology of the region.

Water Quality

The quality of the groundwater in Dade County is intimately related to man's activity. Because of the relatively immature and shallow soils and the permeability of the limestone, water quality is quickly affected by discharge or exfiltration of untreated or improperly treated domestic and industrial wastewater, accidental spillage or leakage of chemicals on or under the ground, and washdown or runoff from equipment storage yards, parking lots and roadways. The potential for degradation also arises from the cumulative impact of many small individual sources, such as the chemicals used in routine lawn maintenance, household chemicals, cleaning and disinfecting agents discharged through septic tank drainage fields, etc. The Biscayne Aquifer is the only source of potable water for this entire metropolitan area. The areas of particular concern are within the cones of influence of the wells from which potable water is drawn and the backwater areas within 300' of a lake or other standing body of water. (CDMP, '79, p. 45-47).

The Local Ecosystems

The local ecosystems vary depending on the elevation, water table, and presence or absence of fire. Exhibit I shows the interrelationship. The uplands or Coastal Ridge area is home to pineland forests and hardwood hammocks. At one time there were over 180,000 acres of pine. Currently, outside the ENP, there are only three pinelands greater than 100 acres. There may be 200 stands of pine of about five acres in size. Periodic fire is an essential component of a pineland, or the area will change to hardwoods or exotics. (CDMP, '79 p.54) The hardwood hammocks are the other natural ecosystem of the uplands. As few as 50 of these remain, most of which are in public ownership. These rank among the most unusual biotic communities in the continental United States. Trees characteristic of the West Indies, such as Gumbo-Limbo, Lysiloma, Jamaican Dogwood, White Stopper, Madeira Mahogany and Strangler Fig mix with Live Oak, Red Bay, Mulberry and many other hardwood species typical of the temperate zones.

Exotics

The south Florida ecosystems have become vulnerable to the invasion of exotic plant and animal species, several of which have become serious problems. The most notable of the exotic plant species are the Melaleuca, Brazilian Pepper and Casuarina trees. The Melaleuca or "punk tree" is a particular problem in and near the Everglades. The tree is both drought and flood resistant and naturally protected from fire by a thick bark. Cutting induces a more vigorous growth by root crown sprouting. The tree can grow anywhere, but prefers areas that have been disturbed, fallow agricultural fields for example. Brazilian pepper and Australian Pine (Casuarina) are similar problem trees. Brazilian Pepper grows in very dense clusters, crowding out almost all other plant life. The sap is irritating to many people and the fruit is poisonous. A mature stand will very seldom burn, and cutting increases propagation. The plant is almost immune to herbicides. Australian Pine is a greater threat to the beach areas than to the Everglades, but it too is a major problem in areas where it has excluded native vegetation and eliminated habitat areas for nesting birds, sea turtles and crocodiles. (CDMP, '79, p. 58).

The Freshwater Wetlands

The lowest order communities are the periphyton, or blue-green alga mat. The most complicated biosystems are the hammocks. The distribution of biotic communities throughout the Everglades is a function of the soils, elevation, water table levels and periods of inundation. The four major subsystems are: 1) the Cypress Hammocks and wet prairies in the northwest, 2) the extensive Sawgrass Marshes interspersed with deeper sloughs, wet prairies and occasional elevated tree islands of the central segment, 3) the Rocky Glades, which generally lie between the central Everglades basin to the northwest and the coastal marsh region of the ENP to the south, and 4) the Coastal Marls or Marl Glades. The Rocky Glades are land that is characterized by rough, rocky outcroppings of limestone known as pinnacle rock. (CDMP, '79, p. 59-62)

VI. ANALYSIS OF VARIABLES

The following paragraphs discuss each listed variable relative to its potential impact on value, the number of sales analyzed, the analysis of the sales, and our conclusions. A summary of the pertinent sales utilized for each factor's analysis is included in the respective sections. A summary description of all the sales is located in the addenda, and the actual data is reproduced on the accompanying CD.

AGRICULTURAL LAND USE

Explanation of Potential Impact: One of the few economically viable land uses for the Florida freshwater wetlands has been the growing of crops and groves for Lechee, Avocado, Citrus Fruit and ornamental plants. Both the Overlay Ordinance and the water management regulations permit continued use of agricultural lands in most instances. Consequently, it is reasonable to expect that the market for land that has been used for agricultural purposes will be different than the market for wetlands intended for passive recreational uses.

Number and Types of Sales Analyzed: A total of 84 sales were coded for agricultural land use based on the historical use at the time of the sale, the stated intended use according to parties to the sale, or the appraiser's comments about highest and best use. Most of the sales that are economically viable for agricultural use are the larger sales. The database contains only two <3 acre sales and eight 5 acre sales that are identified as agricultural use sales.

Statistical Analysis: The multi-linear regression model shows that agricultural land use is a statistically significant explanatory variable (p = 0.0000).

Matched Pairs: Table 8, which follows, partitions the >3 acre sales by size group into agricultural land use sales and non-agricultural land use sales. The table clearly illustrates that the agricultural land use parcels sold a