Introduction & Summary

Following is an exploratory analysis of Total Phosphorus (TP) and flow data collected at sites within and upstream of Everglades National Park (ENP). Output includes maps, figures, and tables reflecting variations associated with location, flow, season, and long-term trends. Results provide a basis for discussions at Everglades Technical Oversight Committee (TOC) meetings regarding factors influencing compliance with Settlement Agreement Limits for ENP inflow TP concentrations and design of monitoring programs for evaluating compliance under future water-management regimes. The analytical framework can also be applied to identify gaps and redundancies in the data for consideration in optimizing the monitoring network.

This birds-eye view of regional data provides basis for comparison with other statistical analyses being performed by the TOC and results described in the South Florida Environmental Reports. Patterns identified in the exploratory analysis can be further explored and interpreted. The software and database interfaces developed to support the analysis can be modified to examine other parameters, sites, time periods, etc.

The analysis is based upon TP & flow data collected by SFWMD at 56 sites in Water Years 2003-2014 (May 2002 - April 2014) at locations representing the inflows, marsh, and outflows of Water Conservation Area 3A (WCA-3A) and ENP. Another report provides a Google Earth interface linked to charts of Flow, TP concentration, and TP load data from most of the long-term monitoring sites in the Everglades basin below Lake Okeechobee.

Results indicate that there were significant decreasing trends in TP concentration at most sites. The trends are consistent with implementation of measures that reduced TP loads to WCA-3A, including startup of STA-34 in 2004, expansion and improving performance of STA-5/6 in 2007-2014, and long-term responses to continued operation of the other STAs and implementation of source controls throughout the basin. The trends were more pronounced at northern sites that had the highest TP concentrations and are closer to WCA-3A inflows (-5 to -10%/year) as compared with interior marsh and southern structures discharging into ENP (-1 to -3%/year). This spatial pattern is likely to reflect dilution by rainfall and release of TP previously stored in the soils, vegetation, and canals of WCA-3A (Walker & Kadlec, 2011). Decreasing trends were typically more prounced during the wet season (May-October) as compared with the dry season (November-April). No trends were indicated at marsh sites in central and southern ENP, which had the lowest TP concentrations and are relatively remote from anthropogenic P sources.

In contrast to TP, there were no significant trends in flow in the agreegated outflows from WCA-3A or at most of the individual structures. It is unlikely that the observed decreasing trends in TP concentration reflected trends in hydrologic conditions driven by climate, as opposed to reductions in TP load to WCA-3A. Trends in flow at some structures can be linked to (a) expansion of STA-5/6 and other measures that reduced flows to the northwest corner of WCA-3A (decreases at G155, S190, S140, S8), (b) startup and increased operation of pump stations during the period (increases at S9A, S332B, S332C), and changes in structure operation at the inflows to ENP Shark River Slough (increase at S333).


All data were downloaded from DBHYDRO. While weekly composite samples were collected at many sites, the analysis was based exclusively on grab samples to provide a consistent basis for spatial analysis. Flow data were retrieved from DBHYDRO using MOD1 or Preferred DBKEYS when available; otherwise source DBKEYS were used. For purposes of summarizing results, sites were assigned to categories reflecting region (WCA-3A/B (inflow, marsh), ENP Shark River Slough (West & East of L67), ENP C111/Taylor Slough) and station type (Structure vs. Marsh). Sample counts by site, water year, and flow category are shown in the attached figure.

Spatial and temporal variations under different hydrologic conditions were analyzed by partitioning the dataset based upon season (wet vs. dry) and flow (all data vs. positive flow). Pairing of water quality sites with flow monitoring sites is indicated in the results summary table. This is straightforward at structure sites. Because flow was not monitored at marsh sites, composite inflows from upstream or downstream structures in the marsh region were used as approximate surrogates for purposes of this exploratory analysis. The estimated flows were used to classify samples into no-flow vs. positive-flow categories; flow magnitudes are not used in analyzing trends in geometric mean TP concentrations at marsh sites. For example, a sample at marsh site NE1 was included in the positive-flow dataset if there was a net positive discharge into Northeast Shark River Slough based upon the measured L29 structure flows (S333 - S334).

The marsh flow classifications are very approximate because of time lags and storage in the marsh, but are sufficient for purposes of this exploratory analysis. Differences between the all-flow and positive-flow results are more evident at the structure sites than at the marsh sites. Classification of marsh data based upon water depths to distinguish wet from dry conditions could be explored in the future, but would be hampered by incomplete data. The wet-season (May-October) vs. dry-season (November-April) comparisons are equally applicable to both structure and marsh sites.

The primary dataset used to evaluate spatial and temporal variations included sampling data collected regardless of flow rate (i.e. includes samples with zero flow). The data were aggregated to monthly averages using the geometric mean for marsh stations and the flow weighted mean for structure stations. Because the structure values were flow-weighted, samples collected on days without flow do not influence the results.

In addition to the primary dataset, the TP spatial and trend analyses were repeated for three alternative datasets to assess sensitivity to the monthly aggregation statistic (geometric mean vs. flow-weighted mean) and to the inclusion of samples with zero measured/estimated flow rate.

For each dataset, the analysis was applied to three seasonal subsets:

The multiple datasets provide a sensitivity analysis of assumptions made in selecting and summarizing the data. As indicated in the examples below, results are displayed in box plots, bar charts, maps, and time series.

Trends in mean monthly flow were also evaluated at each station using the Seasonal Kendall test.


Data displays and statistical analyses were performed on log10-transformed TP concentrations (ppb). Monthly and Water Year (May-April) geometric means were computed from means of log10-transformed values (\(Geomean = 10^{mean(log_{10} TP)}\)). Long-term geometric means listed in the results summary table were computed as arithmetic means of yearly geometric means (algorithm used in FDEP’s 4-Part Test for compliance with the Everglades P criterion). Flow-weighted means were computed from grab-sample concentrations and flows measured on sampling dates.

Trend slopes were computed using the Sen slope estimator, which is the median of all pairwise slopes between each pair of data points in the timeseries. For the log10-transformed TP concentrations, trend slopes are expressed as a percent change in concentration per year (rather than an absolute change in concentration). The percent slope was estimated using the following formula: \(100 \times (10^{Slope(log_{10} TP / yr)} - 1)\). For the monthly mean flow rates, the trend slopes are expressed as a percentage of the long-term mean flow rate at each site.

R scripts developed for this analysis could be adapted for analysis of spatial and temporal variations in other subsets of regional water quality data (parameters, sites, seasons, years, etc…). The scripts could also be modified to address specific questions or hypotheses that evolve from this initial exploratory analysis.


TP Concentrations

TP Sample Counts by Site, Flow Category, & Water Year (csv)

Table of Basic Statistics, Trend Slopes & Significance for TP (xls)

Results by Dataset:
Primary (Marsh Geomeans and Structure Flow-weighted Means) Spatial Variations Trend Maps Trend Analysis Summary
Monthly Geometric Means, All Flows Trend Maps Trend Analysis & Diagnostics
Monthly Geometric Means, Positive Flows Trend Maps Trend Analysis & Diagnostics
Monthly Flow-Weighted Means Trend Maps Trend Analysis & Diagnostics

Appendix: Examples

WQ Monitoring Sites & Categories