For several years, Keeping Forests worked with researchers and partners across the South to ask whether forest landowners could be paid for the benefits their forests and management provide to water. This page brings together what the research and pilot projects established, where the evidence stops, and how the findings created a significant pivot in our approach.
Laura Calandrella, Executive Director Keeping Forests | September 2026
01 — Why water
We wanted to create new income pathways for forest landowners.
In 2019, Keeping Forests began asking whether Southern forest landowners could be paid for benefits their forests already provide but traditional commodity markets do not. Water became the first test.
The premise
Keeping Forests exists to create the economic conditions that enable the South's 245 million acres of forest to remain forested. In 2019, a year-long process with partners across the region identified three high-leverage pathways for doing that. One became our “Emerging Markets” strategy.
Why water
A team of Keeping Forests partners helped define that strategy and the research behind it. We asked whether forest landowners could be compensated for benefits associated with active forest management that traditional commodity markets do not pay for. Water was the first opportunity we pursued. The connection between forests and water was widely recognized, the potential beneficiaries could be identified, and existing payment-for-ecosystem-services models offered a possible way to connect that value back to landowners.
How we work
While many of our partners deliver programs and outcomes on the ground, Keeping Forests works further upstream, on the market architecture that makes those efforts possible. We bring partners together to examine the systems, incentives and sources of capital shaping land-use decisions across the South—and to develop approaches that can be adapted across a region where 86% of forestland is privately owned.
The test
In this case, that meant testing the entire value proposition. We looked at not only what forests provide to water, but whether that benefit could support a workable payment to landowners. Water remains an important part of our work, and we continue to see potential in places where the ecological, economic and local conditions align. At the same time, this work gave us our clearest view yet of what it takes to turn an ecosystem benefit into an actual payment and where that connection becomes difficult.
What's next
We are grateful for the time, expertise and resources our partners invested in this work. We look forward to continued collaboration as we carry what we learned from water into a broader question about the capital already reshaping the South.
— 1 · The premise
We tested the full path from forest management to landowner payment.
Forests affect water quality and the way water moves through a landscape. Utilities, companies and communities depend on that water. The premise seemed straightforward: show how forest management benefits water, measure the value of that benefit, identify who receives it, and create a way for some of that value to reach landowners. The research was designed to examine each of these aspects, while also probing at the assumption that, if ecological and economic value could be demonstrated, a viable market opportunity could follow. The research helped us to understand the nuances of payment-for-ecosystem-services. This is what we learned.
FIGURE 1 — THE RESEARCH WE SET OUT TO ESTABLISH
Impact of active forest management on water
How forested watersheds change water quality and flow
The economic value of forested watersheds to water treatment costs
Whether economic data could motivate people to pay for forest management
Mechanisms to establish a payment-for-ecosystem services water market
We conducted simultaneous peer-reviewed science and field research in each area from 2020-2024.
The assumption
If we could define the economic value that forests provide to water and establish a viable payment mechanism, we could incentivize payment for the ecosystem service.
What we found
Forests’ effects on water varied by place, scale and type of land-use change. Even where the benefits were clear, utility payments were constrained by economics, institutional capacity and public acceptance. Scalable solutions will require starting with the needs and investments already present in a place, then determining how forests and forest landowners can contribute.
— 2 · How the research was built
If forests benefit water, would someone pay landowners for it?
We combined peer-reviewed science with pilot site testing to uncover five different kinds of evidence:
Ecology asked how forests actually impact water.
Economics asked whether those changes can be verified as net-positive benefits to water treatment costs.
Hydrology asked how forests change the way water moves through a watershed.
Land use change modeling asked where forest conversion negatively impacted drinking water.
And, finally, social science asked the pivotal questions surrounding what motivates people and institutions to pay for or actively manage for water quality and quantity benefits as they relate to forested watersheds.
Partners were actively engaged in shaping five lines of questioning
The scientific research
Peer-reviewed studies testing whether the service is real.
Caldwell et al., 2023 · peer-reviewed · Keeping Forests Emerging Markets team acknowledged.
Landowner and pilot work testing whether anyone would, or could, pay.
Dovetail Partners, Georgia Forestry Foundation and TBL Consultants, 2021; Conservation Investment Management Mobile Bay Analysis 2023; Saluda River Basin community-based social marketing pilot with Impact by Design, Lauren Watkins Consulting, and the South Carolina Forestry Commission, 2024 · Keeping Forests–commissioned research and pilot work.
These inquiries complement one another and were, at times, conducted simultaneously. They build on decades of forest-water and payment-for-ecosystem-services research that came before it. Keeping Forests helped assemble, extend and interpret a larger body of work.
— 3 · What the science established
The link between forests and water quality is real. But a case for local investment requires more evidence.
Read together, the research supports a broad conclusion and a clear limit. Across the South, more forest upstream is generally associated with cleaner source water, while conversion to development or agriculture tends to increase nutrients and sediment. But these are patterns across watersheds and water systems—not promises about what will happen on one property or at one drinking-water intake. Caldwell et al., 2023
1,746
drinking-water intakes included in the regional analysis
15%
met the study's thresholds for high-quality source water
Less than 10%
would reduce treatment costs for most water systems
— Reader aid
A guide to water terminology
Water can reach a stream quickly as runoff or slowly through the ground. The amount, timing and quality of that water can all change when land use changes. Select a term to read its definition.
Figure 2 — where the water terms sit in a landscapeWhere the four flow terms sit in a landscape: rain either runs off the surface, or soaks in and returns slowly to the stream that a drinking-water system draws from.
Source water
The untreated water that reaches a drinking-water system. The research looked both at what is in that water and at how water moves through the landscape.
Nitrogen & phosphorus
Nutrients that can cause algae and water-quality problems when too much enters the water.
Sediment
Soil and other material carried into streams.
Turbidity
How cloudy the water is.
Organic carbon
Natural material from plants and soil that can make water more expensive to treat.
Runoff
Rainwater that moves quickly across the land and into streams.
Recharge
Rainwater that soaks into the ground.
Baseflow
Groundwater that slowly feeds streams between storms.
Finding 01
Forest cover matters— but water is ultimately impacted by multiple factors of the natural geography.
More upstream forest is generally associated with cleaner source water. But water quality also reflects soils, geology, precipitation, reservoirs and surrounding land uses. Even in forest-dominated watersheds, some nutrients and sediment come from other natural background conditions. Keeping one tract forested does not guarantee a measurable change downstream.
Share of the variation explained by upstream land cover (R², observed data)
Nitrogen41%
Phosphorus31%
Sediment5%
Land cover is one influence among several. Soils, geology, precipitation, reservoirs and surrounding land use carry also create impact on water quality. Although we can say at a regional scale that forested watersheds have benefits for clean drinking water, that pattern doesn't accurately describe what is happening at one intake.
FIGURE 4 — ECONOMIC IMPACT ON WATER TREATMENT COST BASED ON PROJECTED LOSS OF FOREST
Modest cost effect — most facilities14 facilities over $100k a year
At most of the survey facilities, there was only a modest cost impact that could be demonstrated. Average modeled impact was about $19,000 per facility per year, but 29 facilities exceeded $50,000 and 14 exceeded $100,000. NOTE: This is a schematic diagram for illustrative purposes.
Finding 02
Losing forest can raise treatment costs, but in uneven ways across the region.
Some changes in source-water quality are associated with higher treatment costs. The strongest observed relationship was with organic carbon; modeled changes in nitrogen and phosphorus also affected costs. For most facilities, the modeled effect of future land-use change was modest. A smaller number showed much greater exposure. There is no single dollar value for an acre of forest.
In an extreme modeled scenario, replacing forest with urban development sent much more rain quickly across the surface and into streams. Less water soaked into the ground or was released gradually between storms. The result was not simply “more water.” The scenario demonstrated that this change in land use produced faster runoff and less of the slow, steady flow that helps regulate a watershed. Although the research does not offer precise forecasts, it did reveal the direction and possible scale of change.
FIGURE 5 — THE SAME RAIN TAKES A DIFFERENT PATH AFTER DEVELOPMENT
Forested land
More water soaks in
Stored and released over time
After development
More water runs off
More reaches streams quickly after rain
+85%
surface runoff
−26%
water soaking down through the soil
−22%
slow, groundwater-fed streamflow
More water reaches streams quickly, especially after storms, and less is stored afterward.
FIGURE 6 — HIGH-POLLUTION DAYS COULD BECOME MUCH MORE COMMON
Baseline
About 1 in 10 days
Modeled future — nitrogen
As many as 1 in 3 days
Modeled future — sediment
As many as 2 in 3 days
These are modeled examples from the most affected intake for each pollutant—not results for every water system. They describe untreated source water before treatment and are scenarios, not forecasts.
Finding 04
Forest loss could mean more difficult days for water treatment.
In the Middle Chattahoochee, the model found that forest loss and development could make high-pollution days more common at many drinking-water intakes. On those days, it was hard to manage quality before treatment. Some intakes were affected much more than others, reinforcing the need to understand conditions around each water system.
The economic analysis produced a result that does not fit neatly into a simple “more forest equals lower costs” story. Across the region, the modeled land-use scenarios produced an estimated $7–25 million in annual treatment-cost savings—but those savings were driven largely by projected conversion of agricultural land, not by forest retention.
The ultimate impact on water quality and quantity regionally depends on the full mix of land-use changes, and each change affects nutrients differently in different places.
The authors concluded that avoided treatment costs alone may not be sufficient to justify retaining forest where land values and other opportunity costs are high.
Regional evidence tells us where to look, but it will be the local analysis that tells us whether an investment case exists.
The research is better suited to identifying landscapes where forest retention matters than to assigning a water value to individual acres. That does not rule out payment. It changes the scale at which the opportunity should be considered. The evidence points toward watershed- and landscape-scale strategies in places where forest loss creates a meaningful shared risk—not toward a standard per-acre price based on a promised downstream result. Local evidence is still needed to determine which lands matter, who needs to participate and what kind of investment could work.
FIGURE 7 — REGIONAL EVIDENCE TELLS US WHERE TO LOOK. A LOCAL CASE TELLS US WHETHER TO INVEST.
Regional evidence
Where might forests matter to water?
—Identifies broad patterns
—Highlights places for a closer look
—Shows where forest loss may create water risks
Local evidence still needed
Local investment case
What would make investment worthwhile here?
—Which forested lands matter?
—What change or landowner action is involved?
—Who would experience the result?
—Would it matter enough to change a decision?
Regional research can narrow the search. It cannot make the local investment case by itself.
— 4 · From research to practice
We needed to understand how real landowners, communities, and potential buyers thought about water.
A critical piece of success was to determine how to motivate buyers and sellers to participate in the market. Keeping Forests and its partners pursued in-depth social science with both groups through landowner engagement in Georgia and pilot efforts in the Mobile Bay and Saluda River Basins.
FIGURE 8 — GEORGIA · ALABAMA · SOUTH CAROLINA: THE THREE PLACES WE TESTED
Select a location on the map to learn the key question we answered in each geography.
The question
GEORGIA · FOUR WATERSHEDS
Will landowners participate in a payment-for-ecosystem service market? And, if so, under what conditions?
What we tested here — landowner participation
In 2021, Keeping Forests, the Georgia Forestry Foundation and partners engaged landowners and other stakeholders across four Georgia watersheds through nine focus groups and forums.
We asked what landowners valued, what kinds of payments or incentives interested them, who they would trust to administer a program and what would keep them from participating.
The responses showed interest, but also clear conditions. Participants preferred straightforward agreements, direct payments or tax incentives, and administration by a trusted nonprofit or similar organization. Complicated agreements and restrictions on future timber harvests raised concern.
The question
Mobile Bay, Alabama
Does the potential degradation of water quality and quantity motivate buyers to consider investment in forestland?
What we tested here — buyer motivation
In Mobile Bay, we explored whether downstream beneficiaries might pay landowners for maintaining or improving water quality. The case for utility funding did not hold. The expected water-quality degradation—and the resulting effect on treatment costs—was not large enough to give utilities a compelling reason to pay.
Corporate buyers appeared to offer another possible path, particularly companies with water sustainability goals. But that path was not well developed. Companies generally had a clearer understanding of investments within their own facilities—such as reducing water use or improving wastewater treatment—than of investments elsewhere in the watershed.
Mobile Bay showed us the need for being able to translate a company's broad water commitments into a credible investment in a particular watershed.
The question
Saluda River Basin, South Carolina
How do we connect buyers and sellers? How do we remove barriers to participation?
What we tested here — local action & delivery
In the Saluda River Basin, Keeping Forests worked with the South Carolina Forestry Commission and local partners to examine what would motivate participation on both sides. We used community-based social marketing, or CBSM, which starts with the actions people need to take and identifies what would encourage or prevent them.
For landowners, trust mattered. Consulting foresters, associations, Extension agents and local land trusts already had relationships that a new program could not manufacture. The practical opportunity was to strengthen those organizations, not compete with them. Potential buyers needed a defined action or "shovel-ready" project and evidence they considered credible.
The work produced two practical tools. The first was a database mapping landowner associations, the services they provide, and the landowners they served. The second was a decision-support tool to score and rank companies based on their likelihood of investing in a water- or forest-related project. Together, these two tools offered a repeatable way of examining which buyers and sellers in a watershed that were best suited to initiate an investment conversation.
Four lessons that clarified the starting point for future models.
1
Landowners cannot be treated as the last step.
Payment-for-ecosystem service models often define the criteria for participation and then enroll landowners in their programs. The most scalable solutions will include landowners early in the process to gain their input, trust, and interest in participation.
2
A beneficiary is not automatically a buyer.
One of the biggest challenges for water markets is that the beneficiaries who could be buyers are diffuse. Just because an individual or institution depends on water
doesn't mean they have motivation, authority or capacity to pay for upstream forest outcomes.
3
Start with the relationships already in place.
Landowner associations, consulting foresters, Extension staff and other local organizations already connect with landowners. A new model should strengthen those relationships rather than attempt to replace them.
4
Choose the structure last.
Any solution, whether a water fund, conservation easement or carbon credit, may be a powerful solution in one watershed and not another. The people, institutions and source of payment have to be understood before the structure is chosen.
Keeping Forests' Take
We need a scalable way to get capital to landowners without turning every forest benefit into a new credit or program.
Carbon markets were built to scale by turning different forest projects into standardized, tradable units. Many efforts to create water markets have followed a similar path: define the benefit, create the product and then recruit buyers and landowners. These approaches have produced real transactions, but their uneven results also show what standardization cannot solve—whether the benefit matters enough to a buyer, whether landowners will participate and whether the model fits the place.
Many mature markets evaluate and structure transactions one at a time. They scale because the process, standards and supporting infrastructure can be repeated. Our aim is not to force every place into the same payment model. It is to build a repeatable way to connect local forest conditions, landowners and capital—and, over time, assemble those individual agreements into something larger.
— 5 · Evidence Explorer
What the details of the research say and where they hit their limits.
The Evidence Explorer translates the key findings from all of our peer-reviewed science and field work. We present the findings in a question and answer format to make it easy to understand, but it's important to note that some of the research was geography-specific. We have identified where that is true. For those who want to go deeper, the full evidence record shows the connected study and the data citation we are drawing from.
* "Southeastern U.S." refers to the 13 Southern states in these studies: Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, and Virginia.
6 of 49 records showing
What it supports — the number, verbatim
All regressions of TN, TP, TSS on % upstream forest were significant and negative; developed/ag/other-natural generally significant and positive (one exception: TSS vs developed n.s., p=0.21).
Type & scale
STATISTICAL ASSOCIATION (observed) · Watershed / catchment (% of upstream drainage area)
What it does not tell us
Does not establish a causal per-acre effect or a site-specific water value.
Key limitation
Large site-to-site variability for any given % forest.
Applies directly to one property? No — regional cross-section, not per-parcel
Can it be mapped? Yes
Keeping Forests' relationship
KF Emerging Markets team acknowledged in this paper.
Source
Caldwell et al. 2023 · Science of the Total Environment 882:163550
"a 1% change from forest to developed land cover could result in an approximately 1.5 +/- 0.2% increase in TN, a 1.9 +/- 0.3% increase in TP, and a 0.4 +/- 0.5% increase in SS." (SS coefficient not significant, p=0.13.)
Type & scale
DERIVED / STATISTICAL ASSOCIATION · Regional mean elasticity across monitoring sites
What it does not tell us
Not a per-parcel or PES pricing figure; the +0.4% sediment result crosses zero (not significant).
Key limitation
+/- 95% CI must be kept; SS effect not significant.
Applies directly to one property? No — a regional mean, not a tract prediction
Can it be mapped? Partly
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Caldwell et al. 2023 · Science of the Total Environment 882:163550
"a 1% change from forest to agricultural land cover could result in an approximately 2.4 +/- 0.3% increase in TN, a 3.2 +/- 0.4% increase in TP, and 1.4 +/- 0.7% increase in SS."
Type & scale
DERIVED / STATISTICAL ASSOCIATION · Regional mean elasticity
What it does not tell us
Same regional-mean caveat as E2.
Key limitation
Wide CIs, esp. sediment.
Applies directly to one property? No
Can it be mapped? Partly
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Caldwell et al. 2023 · Science of the Total Environment 882:163550
"Catchments with dominant (>90%) agricultural land cover upstream had the greatest export rates for all parameters, followed by developed, then forest and other-natural."
Type & scale
MODELED (SPARROW) · Catchments >90% single cover (16,139)
What it does not tell us
A >90% dominance threshold, not a marginal-acre value.
Key limitation
Modeled export, not measured.
Applies directly to one property? No
Can it be mapped? Yes
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Caldwell et al. 2023 · Science of the Total Environment 882:163550
"the 90th percentile TN export from forest-dominant catchments (4.2 kg ha-1 yr-1) may be greater than the 10th percentile TN export from developed-dominant catchments (3.2)."
Type & scale
MODELED · Catchment distributions
What it does not tell us
Undercuts any claim that a forested parcel always yields cleaner water than a developed one.
Key limitation
Driven by natural background sources; this is the paper's central caveat.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Caldwell et al. 2023 · Science of the Total Environment 882:163550
"atmospheric N deposition... contributing 90% or more of the TN export for >90% forest-dominant" catchments; geologic P >=50% of TP export in forest-dominant catchments.
Type & scale
MODELED · Forest-dominant catchments
What it does not tell us
A large share of outcomes in forested watersheds is not attributable to protection at all — relevant to avoided-loss crediting.
Key limitation
Modeled source apportionment.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Caldwell et al. 2023 · Science of the Total Environment 882:163550
"median estimated TN concentration was 1.7 times greater for run-of-river intakes with <50% forest upstream but 1.3 times greater for intakes on reservoirs with <50% forest."
"Of all PWS intakes, 15% had high raw water quality, and 85% of those were on reservoirs. Of the run-of-river intakes with high raw water quality, 75% had at least 50% forest land cover upstream."
Type & scale
MODELED + DERIVED · 1,746 intakes
What it does not tell us
Forest is associated with, not proven sufficient for, high quality.
Key limitation
Threshold-based classification.
Applies directly to one property? No
Can it be mapped? Yes
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Caldwell et al. 2023 · Science of the Total Environment 882:163550
"98,658 km2 (7.3%) of land in natural land cover in 2020... predicted to be converted to developed land by 2070"; smaller watersheds saw the largest projected reductions.
"a 1% increase in nitrogen concentration leads to an increase in costs... by 0.76% and 0.91%... a 1% increase in phosphorus increases the unit costs by 0.37% and 0.46%." (P weaker/mixed significance.)
Type & scale
DERIVED (simulated water quality) · Facility-level (39 obs)
What it does not tell us
Based on simulated, not measured, nutrient data; not a payment-ready value.
Key limitation
Simulated concentrations; small sample.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Nehra et al. 2025 · Forest Policy and Economics 179:103603
"a 1% change in forest loss in an upstream watershed results in a roughly 1.7% increase in treatment costs" (TN model; coefficients -1.7344 to -1.7142 across four scenarios).
"For TP, a 1% loss of upstream forest cover increases treatment costs by roughly 1%... the benefit of increasing upstream forest cover is lower by about 0.7 percentage points."
Type & scale
PROJECTED / MODELED · ~1,359 facilities
What it does not tell us
Asymmetry: forest loss costs more than forest gain saves.
Key limitation
Model-dependent.
Applies directly to one property? No
Can it be mapped? Yes
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Nehra et al. 2025 · Forest Policy and Economics 179:103603
"total projected regional cost changes are negative... ranging from $7-$25 million yr-1 in cost savings, but with most of these savings occurring in areas with high relative levels of agricultural land conversion."
Type & scale
PROJECTED / ECONOMIC ESTIMATE · Region-wide net
What it does not tell us
Does NOT support 'forests save $X region-wide'; the net is dominated by agricultural change and can run against the forest thesis.
Key limitation
The regional net is driven mostly by projected cropland conversion, not by forest protection — it should not be read as 'forests save money region-wide.'
Applies directly to one property? No
Can it be mapped? Partly
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Nehra et al. 2025 · Forest Policy and Economics 179:103603
"~$7 million per year... average cost impact ~$19,000 per year, but 29 facilities show >$50,000 per year and 14 facilities >$100,000 per year" (one scenario, TN, 0-10% band).
Survey to 1,300 utilities, response rate "less than 5%"; final 37 surveys / 32-39 usable observations; authors ran extra uncertainty analysis "due to the limited number of observations."
Type & scale
STATED LIMITATION · 32-39 observations
What it does not tell us
The economics is screening evidence for exposure, not a transaction-grade appraisal.
Key limitation
Small non-random sample; underrepresents small rural utilities.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Nehra et al. 2025 · Forest Policy and Economics 179:103603
"avoided drinking water treatment costs from forestland preservation may not be high enough to justify land preservation exclusively for source water quality protection in areas with high opportunity costs."
Type & scale
AUTHOR CONCLUSION · n/a
What it does not tell us
It does not say forests aren't worth protecting — only that water-treatment savings alone may not pay for it where land is valuable.
Key limitation
An author judgment drawn from the study's results, not a separate measurement.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF Emerging Markets team acknowledged.
Source
Nehra et al. 2025 · Forest Policy and Economics 179:103603
"current land use in 2020 may have increased the annual average sediment loads by 13%-80% and nutrient loads by 14%-40% relative to natural land uses."
Type & scale
MODELED (scenario contrast) · Major tributaries
What it does not tell us
A model contrast, not an observed historical trend.
"projected land use change could increase annual average river discharge by as much as 6%, sediment load by ~40%, total nitrogen load by ~10% and total phosphorous load by ~20% across major tributaries."
"Lake Conroe could experience a 6% increase in annual average streamflow and a 10% increase in sediment load, reducing reservoir storage capacity over time."
Type & scale
PROJECTED / MODELED · Lake Conroe reservoir
What it does not tell us
Reservoir sedimentation dynamics not directly modeled (future work).
"Uncertainties in model inputs, structure, parameters, and future land use estimates can compound"; SWAT "unable to accurately capture peak TSS loads."
Type & scale
STATED LIMITATION · n/a
What it does not tell us
Directional insight into mechanism — not a quantitative forecast for any location.
Key limitation
Point sources held constant; peak sediment underestimated.
"60.5% of the watersheds used for drinking water in the state are forested"; forests ~58% of land use (~22M acres).
Type & scale
SECONDARY / SYNTHESIS · State
What it does not tell us
A secondary summary, not primary research — cite as Dovetail/GFF 2021, kept distinct from the papers.
Key limitation
Literature review, not new data.
Applies directly to one property? No
Can it be mapped? Yes
Keeping Forests' relationship
Co-commissioned BY Keeping Forests (with Georgia Forestry Foundation).
Source
Dovetail Partners, Georgia Forestry Foundation & TBL Consultants, 2021
Dovetail/GFF 2021 (report)
What it supports — the number, verbatim
"more than 70% of family forest owners in Georgia indicated 'protecting water resources' as an important reason for owning forestland" (NWOS).
Type & scale
SECONDARY (NWOS survey) · Landowner population
What it does not tell us
Attitude, not behavior; national survey applied to Georgia.
Key limitation
Secondary survey data.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF-commissioned report.
Source
Dovetail Partners, Georgia Forestry Foundation & TBL Consultants, 2021
Dovetail/GFF 2021
What it supports — the number, verbatim
Woodland Retreat 41%, Working the Land 28%, Supplemental Income 19%, Uninvolved 12%; "89%... classified as 'Prime Prospects'."
Type & scale
SOCIAL SCIENCE (segmentation) · Landowner segments
What it does not tell us
Segmentation model, not a census; behavior may differ from attitude.
Key limitation
TELE/SFFI framework on NWOS data.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF-commissioned report.
Source
Dovetail Partners, Georgia Forestry Foundation & TBL Consultants, 2021
Dovetail/GFF 2021
What it supports — the number, verbatim
Preferred administrator: "1. Nonprofit/NGO"; preferred payment: "1. Direct payments 2. Tax incentives"; top concern: "overcomplicated agreements, restrictions on harvests."
Type & scale
STAKEHOLDER DATA (forums) · ~350 forum registrants
What it does not tell us
Authors call it "a small and non-scientific sampling process" — present as receptivity, not a survey.
Key limitation
Non-scientific engagement sample.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF-commissioned report.
Source
Dovetail Partners, Georgia Forestry Foundation & TBL Consultants, 2021
Dovetail/GFF 2021
What it supports — the number, verbatim
"2 million acres of forest in Georgia have the potential to be lost through conversion... between 2030 and 2060, primarily due to urban growth."
Type & scale
SECONDARY / PROJECTION · State
What it does not tell us
Georgia-specific projection cited from secondary sources.
Key limitation
Projection.
Applies directly to one property? No
Can it be mapped? Yes
Keeping Forests' relationship
KF-commissioned report.
Source
Dovetail Partners, Georgia Forestry Foundation & TBL Consultants, 2021
Dovetail/GFF 2021
What it supports — the number, verbatim
Local utilities interviewed n=1 despite many attempts ("we tried hard for more"); local landowner associations n=0. Utility quote: "There's no way we can add a new fee... it'll just seem like another tax."
Type & scale
PILOT / KEY-INFORMANT INTERVIEW (qualitative) · One basin / single utility
What it does not tell us
How utilities behave in urban systems with a large ratepayer base and admin capacity.
Key limitation
Single interview; rural-cooperative context; qualitative.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF-led pilot (with Impact by Design / L. Watkins).
Source
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Impact by Design CBSM Final Report 2024; Watkins webinar Aug 2024
What it supports — the number, verbatim
Online panel survey n=81 (Laurens County, SC): substantial share on private wells; small-cooperative service territory.
Type & scale
PILOT SURVEY (non-representative) · Single county
What it does not tell us
Behavior or region-wide representativeness (small, single-county sample).
Key limitation
n=81; receptivity not behavior.
Applies directly to one property? No
Can it be mapped? Partly
Keeping Forests' relationship
KF-led pilot.
Source
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Impact by Design CBSM Final Report 2024; Watkins webinar Aug 2024
Whether a tailored rural model could work; quantified thresholds.
Key limitation
Small expert sample.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF-led pilot.
Source
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Impact by Design CBSM Final Report 2024; Watkins webinar Aug 2024
What it supports — the number, verbatim
Interviews + workshops (consulting foresters n=4+, focus groups n=6, in-person workshop n=14). Forester quote: "We know who to call because we've been doing it long enough."
Type & scale
PILOT INTERVIEWS / WORKSHOPS · Community
What it does not tell us
An exact ranking; trust varies place to place.
Key limitation
Qualitative; one basin.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF-led pilot.
Source
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Impact by Design CBSM Final Report 2024; Watkins webinar Aug 2024
What it supports — the number, verbatim
Qualitative synthesis across pilot + PES enrollment literature. Expert quote: "There are already several groups like this... engagement would be pretty low."
Type & scale
PILOT SYNTHESIS + PES LITERATURE · Program / community
What it does not tell us
A quantified enrollment effect size.
Key limitation
Qualitative; directional.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF-led pilot.
Source
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Impact by Design CBSM Final Report 2024; Watkins webinar Aug 2024
What it supports — the number, verbatim
10 associations, 49 variables; ground-truthed with 4 consulting foresters. Top 3 gaps: online presence; sponsors/partnerships; targeted support.
Type & scale
PILOT DECISION-SUPPORT TOOL (desktop research) · 10 associations
What it does not tell us
A full regional census; whether filling a gap changes outcomes.
Key limitation
Pilot-scale; desktop + 4 forester interviews.
Applies directly to one property? No
Can it be mapped? Partly
Keeping Forests' relationship
KF-led (decision-support tool).
Source
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Impact by Design CBSM Final Report 2024; Watkins webinar Aug 2024
What it supports — the number, verbatim
10 corporations, 46 variables; scored 0-3 per variable, forest/water variables weighted higher; sample profile built for top scorer.
Type & scale
PILOT DECISION-SUPPORT TOOL · 10 corporations
What it does not tell us
Whether ranked firms will actually invest; 'social license' was hard to assess and dropped.
Key limitation
Pilot-scale; desktop research.
Applies directly to one property? No
Can it be mapped? Partly
Keeping Forests' relationship
KF-led (decision-support tool).
Source
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Impact by Design CBSM Final Report 2024; Watkins webinar Aug 2024
What it supports — the number, verbatim
~89% 'prime prospects' (NWOS/TELE segmentation). See also D3/D4.
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Dovetail/GFF 2021
What it supports — the number, verbatim
Named across interviews, workshops, and the association database ('Gaps in Local Landowner Support').
Type & scale
PILOT INTERVIEWS + ASSOCIATION RESEARCH · Community + associations
What it does not tell us
Which barrier dominates in a given place.
Key limitation
Qualitative.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF-led pilot.
Source
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Impact by Design CBSM Final Report 2024; Watkins webinar Aug 2024
What it supports — the number, verbatim
Webinar closing takeaway: "Existing corporate metrics will be the motivator and the measurement to catalyze a market."
Type & scale
PILOT CONCLUSION (2024) · Strategy-level
What it does not tell us
Whether corporate ESG money pays for the water service itself vs. reputation.
Key limitation
PROVENANCE NOTE: where the pilot pointed in 2024; KF's thinking has since moved toward capital with a material dependence on the landscape (see page S7). Kept as honest provenance, not a current position.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF-led pilot; superseded by KF's 2026 read.
Source
Impact by Design CBSM Final Report 2024; Watkins webinar, Aug 2024
Impact by Design CBSM Final Report 2024; Watkins webinar Aug 2024
What it supports — the number, verbatim
TSS rose at 13 of 15 intakes; up to +318% at a small tributary intake (Dog River, 3.1->13.1 mg/L under a high-growth scenario); watershed outlet +9.2% under high growth.
Type & scale
MODELED / PROJECTED (SWAT, 2070 RPA scenarios) · 15 drinking-water intakes; tributary subwatersheds <1,000 km2 most sensitive
What it does not tell us
Not a per-parcel value; a scenario, not a prediction; model underpredicted TSS at some upstream sites.
Key limitation
All new development assumed low-intensity (38% impervious) -> likely conservative; nutrient/sediment error accumulates.
Applies directly to one property? No
Can it be mapped? Yes
Keeping Forests' relationship
KF collaborated on this work (confirmed by KF). The paper carries no explicit printed acknowledgment of Keeping Forests -- credit as KF-involved, not as a formal acknowledgment.
TN rose at 13 of 15 intakes; up to +220%; Lake Harding intake >100% across all future scenarios (0.36->1.2 mg/L); watershed outlet +15% under high growth.
Type & scale
MODELED / PROJECTED (SWAT, 2070 RPA scenarios) · 15 intakes; tributary and smaller subwatersheds most sensitive
What it does not tell us
Not a per-parcel value; TN is the hardest variable to model (errors accumulate from flow -> sediment -> nutrients).
Key limitation
Default fertilizer rates held constant; point-source N loaded as mobile nitrate.
Applies directly to one property? No
Can it be mapped? Yes
Keeping Forests' relationship
KF collaborated on this work (confirmed by KF). The paper carries no explicit printed acknowledgment of Keeping Forests -- credit as KF-involved, not as a formal acknowledgment.
Extreme sediment days 3.6-6.6x more frequent; e.g., Dog River sediment extremes 402 -> 2,667 excess days (6.6x); Lake Harding nitrogen extremes 3.6x (402 -> 1,443 days).
Threshold set by the baseline period only; future climate held constant to isolate land use.
Key limitation
Gridded-climate spatial uncertainty; land use, not climate, is the driver here.
Applies directly to one property? No
Can it be mapped? Yes
Keeping Forests' relationship
KF collaborated on this work (confirmed by KF). The paper carries no explicit printed acknowledgment of Keeping Forests -- credit as KF-involved, not as a formal acknowledgment.
Only 2 of 15 intakes (Snake Creek, Hillabahatchee Creek) saw TSS and TN fall across scenarios; both had agricultural land decline (e.g., 14%->9%) and stable or rising forest.
A minority case (2 of 15); still a modeled scenario, not a plan.
Key limitation
RPA projections downscaled from county scale; limited control of within-subbasin pattern.
Applies directly to one property? No
Can it be mapped? Partly
Keeping Forests' relationship
KF collaborated on this work (confirmed by KF). The paper carries no explicit printed acknowledgment of Keeping Forests -- credit as KF-involved, not as a formal acknowledgment.
Development rises >=50% at 11 of 15 intakes under high growth; the hardest-hit serve small towns (median pop ~3,041) in subwatersheds <1,000 km2; up to 20% upstream forest loss at the most affected intakes.
A projection; assumes mainstem intakes with large drainage areas are less sensitive.
Key limitation
Upper-watershed land use held constant at the model boundary.
Applies directly to one property? No
Can it be mapped? Yes
Keeping Forests' relationship
KF collaborated on this work (confirmed by KF). The paper carries no explicit printed acknowledgment of Keeping Forests -- credit as KF-involved, not as a formal acknowledgment.
Fully developed vs fully forested: water yield +158 mm (+31%), ET -80 mm (-9%); TSS at one intake ranged 126-8,225 metric tons between the forested and developed extremes.
Hypothetical bounds, not scenarios anyone would build; a validation check only.
Key limitation
Confirms response magnitude, not within-subbasin spatial detail.
Applies directly to one property? No
Can it be mapped? No
Keeping Forests' relationship
KF collaborated on this work (confirmed by KF). The paper carries no explicit printed acknowledgment of Keeping Forests -- credit as KF-involved, not as a formal acknowledgment.
10 questions to ask before choosing an investment model
Our water work did not produce one payment model that could be repeated across the South. It did give us a clearer way to determine whether an effort to pay landowners for forest benefits could work in a particular place.
These ten questions cover the full proposition: the benefit, the landowner action, the evidence, who has something at stake, who might pay, who would participate and how an agreement would work. We believe they should be asked before a structure is chosen—not after buyers and landowners are being recruited.
The responses below reflect what this body of water research taught us. They are not universal answers, and they may be different for another place or another forest benefit.
Phase 1
Is there a real benefit?
What the research shows
At a regional and landscape scale, more upstream forest is associated with cleaner source water, and converting forest to development or farmland tends to raise nutrients and sediment and change how water moves through a watershed. The pattern holds across many watersheds and drinking-water intakes.
What must be answered locally
—Which forested area is physically connected to the water source?
—What change or threat is in play here?
—Which water outcome matters in this place?
—Who would actually experience it?
What it does not establish
That keeping any one particular tract forested will produce a measurable benefit for one particular water system. These are patterns across places, not a promise for a single property.
Caldwell et al. 2023 · Gay et al. 2025 · Rath et al. 2026
What the research shows
The best-supported action is avoided conversion: keeping forested land from becoming development or farmland. Some active-management practices may also affect a water variable, but that evidence is thinner, and where it exists the gain is often in water quantity — distinct from the water-quality story.
What must be answered locally
—Is the action avoided conversion or a management practice?
—Is the claim about water quantity or quality?
—What level of proof does this buyer require?
—Who pays to establish it?
What it does not establish
That a specific practice on a specific tract lowered a specific utility's costs. A general benefit is not the same as gallons or water quality delivered to a particular buyer.
Caldwell et al. 2023 · Gay et al. 2025
What the research shows
The research can defensibly describe regional statistical patterns (Caldwell) and modeled watershed scenarios (Rath in the San Jacinto; Gay in the Middle Chattahoochee, under 2070 land-use projections). These are directional scenarios, not forecasts, and land cover alone explains only a minority of water-quality variation — almost none of it for sediment.
What must be answered locally
—Is there monitored, tract-level data tying an action to a measured change?
—What baseline would you measure against?
—What counts as credible proof to the people involved?
What it does not establish
A precise value for any single parcel, or a dated prediction for one intake. The modeling maps possibilities, not certainties.
Caldwell et al. 2023 · Rath et al. 2026 · Gay et al. 2025
Phase 2
Does it matter enough to act?
What the research shows
The research can map who sits downstream of forested source watersheds — the candidate beneficiaries. Caldwell mapped 1,746 drinking-water intakes across the South, with run-of-river intakes the most sensitive to upstream forest; Gay mapped 15 Middle Chattahoochee intakes and projected future risk to them.
What must be answered locally
—What is the physical path from this land to this water user?
—How much of their water depends on the forested area?
—Is the connection direct enough to matter to them?
What it does not establish
That being downstream means having enough at stake to pay. Connection is not value — a beneficiary is not automatically a buyer.
Caldwell et al. 2023 · Gay et al. 2025
What the research shows
Upstream change can carry downstream financial consequences — sometimes material, often not. Nehra modeled roughly a 1.7% treatment-cost change per 1% of forest lost (nitrogen model, ~32 utilities); region-wide the effect nets to a small saving driven mostly by farmland conversion. Gay found projected degradation falls hardest on small-town systems with the least treatment capacity.
What must be answered locally
—What is the actual consequence for this beneficiary?
—Is the financial exposure plausibly material here?
—If not financial, is there a stewardship or reputational reason to act?
What it does not establish
A universal business case or a payment formula. The 1.7% figure is a modeled regional relationship, not a rate you can apply to a place or an acre.
Nehra et al. 2025 · Gay et al. 2025
What the research shows
A beneficiary becomes a buyer only when the benefit is material and the organization has motive, budget and authority — and the evidence a deal needs follows the buyer's reason for paying. In the pilots, Mobile Bay ruled out conventional buyers until only a business genuinely dependent on the resource remained; Saluda's clearest signal was a company asking about shovel-ready work.
What must be answered locally
—What is this buyer's actual reason to pay?
—Do they have the budget and authority to act?
—Can several beneficiaries who value different slices be brought together, instead of seeking one payer for everything?
What it does not establish
That a water goal, a downstream position, or an ESG commitment automatically means willingness to pay. Interest is not a budget.
Mobile Bay pilot · Saluda pilot 2024
Phase 3
Can an agreement work here?
What the research shows
Landowner participation is a behavioral and practical problem, not just a price. The Saluda pilot found trust decisive — consulting foresters, associations, Extension and local land trusts hold relationships a new program can't manufacture. Dovetail found about 89% of family forest owners are willing “prime prospects” who stay unengaged, wary of complex agreements and harvest restrictions.
What must be answered locally
—Who are the trusted messengers here, and are they part of the effort?
—Is the agreement simple enough to say yes to?
—What technical, privacy and succession concerns need addressing?
What it does not establish
That a high enough payment alone will drive participation. Money doesn't overcome distrust, complexity, or a program that competes with the local groups landowners already rely on.
Saluda pilot 2024 · Dovetail 2021
What the research shows
A contracting path can be designed for a specific buyer and place; pooled funds and service agreements scale better than land acquisition. Dovetail found acquisition costly and service agreements most scalable; Mobile Bay pointed toward an outcomes fund; Saluda toward a single water fund brokering between corporations and landowner associations.
What must be answered locally
—Which structure fits this buyer and place?
—Who administers it, and who do landowners trust to?
—Can it pool participants rather than depend on land purchase?
What it does not establish
That a utility water-fund model transfers to a given place, or that a tradable credit market is the answer. The right structure is place- and buyer-specific.
Dovetail 2021 · Mobile Bay pilot · Saluda pilot 2024
What the research shows
Existing third-party standards — FSC, SFI, ATFS, state BMPs — and the data landowners already keep can anchor a low-burden way to verify the action, and corporate accounting frameworks can verify a project-level benefit. The path is to translate what landowners already measure into what buyers need to report — verification without reinvention.
What must be answered locally
—What measurable action is being paid for?
—What baseline and indicators will the buyer accept?
—Can existing certifications or BMP records do the verifying?
What it does not establish
That we can yet credibly verify a specific downstream water outcome tract by tract. Verifying the action is feasible now; verifying the precise outcome is not.
Forest certification & state BMP standards · Dovetail 2021
What the research shows
Across the whole body of work, the recurring blockers aren't ecological — they're institutional: shared benefits with no single owner, avoided-loss returns, no internal decision-maker on the buyer side, and no ready-made transaction to step into. What's needed is one completed, repeatable deal with every earlier question cleared together.
What must be answered locally
—Is there a trusted intermediary to hold the pieces together?
—Is there a decision-maker on the buyer side who can commit?
—Is the structure repeatable, not one-off?
What it does not establish
That good ecology creates a market. A real benefit is necessary but not sufficient; without the institutional pieces, nothing closes.
The full body of work, 2020–2024
What we are carrying forward
Start with the people, the place and the reason to act. Then determine the benefit, evidence and payment structure that fit.
— 7 · Our pivot
We're now aligning the capital investments being made in the region with forest watershed resilience.
We began with a forest-water benefit and looked for someone willing to pay for it. That approach may still work where the water need, utility capacity, local relationships and evidence align. But our work did not uncover a reliable source of water payments at the scale needed to reach landowners across the South.
At the same time, far larger investments are already reshaping the region—in manufacturing, energy, infrastructure and development. Those investments depend on land, affect communities and landscapes, and bring their own needs and pressures.
So we changed the question. Instead of starting with a water benefit and searching for a buyer, we began asking: Where is capital already being committed? What does that investment depend on, and what will it change? Could forests and forest landowners help meet a real need—and could some of that investment reach them in return?
Keeping Forests' Take
Water can be part of the value without being the product.
Water remains part of the equation, but it does not always have to be the product being sold. The ten questions developed through this work give us a way to examine other sources of investment: Is there a real need? Can forests make a meaningful difference? What action would landowners take? What evidence is necessary? Who needs to participate? How could the money reach the land?
We are moving from searching for buyers for a predetermined forest benefit to building agreements around the needs of a particular place and the people investing and living there. The goal has not changed: creating workable ways for landowners to be paid for forest benefits that traditional markets overlook.
The question we're carrying forward
How can forests and forest landowners be part of the South's growth economy?
FIGURE 9 — HOW THE QUESTION CHANGED
We began with
What value do forests provide?
Who benefits?
Will they pay?
We now ask
Where is capital already moving?
What does it depend on or change?
Can forests and landowners be part of the answer?
How can that investment reach the land?
— A podcast by Keeping Forests
How the River Flows
Join Keeping Forests as we talk with our partners, entrepreneurs, and experts about the connection between healthy forests and clean drinking water.
We are grateful to all of the partners who helped bring this research to life.
Keeping Forests did not do this work alone. A working group of partners from across the South helped shape the Emerging Markets strategy, guide the questions and interpret what we were learning.
The research also builds on decades of forest-water and payment-for-ecosystem-services work. Researchers with the USDA Forest Service Southern Research Station, NC State University and collaborating institutions extended that foundation through peer-reviewed studies of land cover, water quality, hydrology and drinking-water treatment costs.
Landowners and local partners made it possible to test the ideas outside the research setting. Landowners in Georgia shared directly what would encourage or prevent participation. Partners in Mobile Bay helped examine whether local water conditions could support buyer investment. In the Saluda River Basin, the South Carolina Forestry Commission and local organizations helped explore what it would take to move buyers, landowners and trusted local partners toward action.
Keeping Forests helped bring these strands together: the broader research, the new scientific work, the landowner perspectives and the place-based pilots. This Evidence Explorer reflects our interpretation of what that collective work established, where its limits remain and what it means for our work going forward.
More than 30 partner organizations and countless individuals contributed to the development of this work.
Funding partners
USDA Forest Service · U.S. Endowment for Forestry & Communities · WWF · Georgia Forestry Foundation · South Carolina Forestry Commission · Georgia-Pacific
Research partners
USDA Forest Service Southern Research Station · NC State University · Texas A&M University
Place-based & landowner work
Georgia Forestry Foundation · Dovetail Partners · TBL Consultants (Georgia) · South Carolina Forestry Commission · Impact by Design / Lauren Watkins Consulting (Saluda River) · Conservation Investment Management (Mobile Bay) · Upstate Forever · Southern Regional Extension Forestry · The Longleaf Alliance · Southern Group of State Foresters · National Alliance of State Foresters · consulting foresters, landowner associations, and community participants
Working group & collaborating partners
The Conservation Fund · Georgia Forestry Association · The Jones Center · South Carolina Forestry Association · South Carolina Rural Water Association · Southeastern Partnership for Forests and Water · Sustainable Forestry Initiative (SFI) · U.S. Fish and Wildlife Service
Research citations
Caldwell, P.V., Martin, K.L., Vose, J.M., Baker, J.S., Warziniack, T.W., Costanza, J.K., Frey, G.E., Nehra, A., Mihiar, C.M. (2023). “Forested watersheds provide the highest water quality among all land cover types, but the benefit of this ecosystem service depends on landscape context.” Science of the Total Environment 882:163550. DOI ↗
Nehra, A., Baker, J.S., Caldwell, P.V., Martin, K.L., Warziniack, T.W., Manner, R.H., Mihiar, C.M., Frey, G.E., Costanza, J.K. (2025). “The potential impact of forest loss on drinking water treatment costs in the southeastern U.S.” Forest Policy and Economics 179:103603. DOI ↗
Gay, et al. (2025). Projected land-use change and drinking-water intakes in the Middle Chattahoochee Watershed. PLOS Water e0000313. DOI ↗
Rath, S., Caldwell, P., Moore, S., Spellman, P., Srinivasan, R., Arnold, J., Martin, K., Sun, G., Moore, G., Vose, J. (2026). “Benefits of Forests for Water: Projected Effects of Land Use Change in the San Jacinto Watershed, Texas.” JAWRA 62:e70083. DOI ↗
Fernholz, K., McFarland, A. (Dovetail Partners) & Klang, J. (TBL Consultants) (2021). “Understanding Payments for Ecosystem Services: Opportunities for Forests, Water and Private Landowners in Georgia and the Southeastern United States.” Prepared for Georgia Forestry Foundation and Keeping Forests. Full report ↗
Impact by Design, Lauren Watkins Consulting and the South Carolina Forestry Commission (2024). Saluda River Basin community-based social marketing pilot — final report and webinar.
