1. IP Protection Framework in Biodiversity Mapping Systems

(A) Copyright (Software + Maps)

Protects:

• Source code of biodiversity algorithms (e.g., species distribution models)
• Visual maps, heatmaps, ecological dashboards

Does NOT protect:

• Raw facts (species presence, coordinates)
• Mathematical methods or ecological theories

(B) Trade Secrets

Protects:

• Training datasets (e.g., rare species occurrence data)
• ML model weights
• Proprietary ecological feature engineering methods

Requires:

• Secrecy
• Commercial value
• Reasonable protection measures

(C) Database Rights (EU-specific)

Protects:

• Substantial investment in collecting biodiversity datasets
• Not the data itself, but the structure/investment

(D) Patent (Limited in practice)

May protect:

• Novel computational methods for ecological prediction
• AI-based habitat classification techniques
  But excluded if considered “abstract algorithms”

2. Key Case Laws Relevant to Algorithmic Biodiversity Mapping

1. Feist Publications v. Rural Telephone Service (1991)

Core Issue:

Whether a simple factual database (telephone directory) can be protected by copyright.

Judgment:

The U.S. Supreme Court ruled:

• Facts are NOT copyrightable
• Only original selection or arrangement is protected

Principle Established:

“Sweat of the brow” (effort alone) is not enough for copyright.

Relevance to Biodiversity Mapping:

• Species occurrence data (latitude/longitude records) = facts → NOT protected
• A biodiversity database is protected only if:
  • It has creative structure (e.g., novel classification system)
• Raw ecological datasets cannot be monopolized via copyright

2. British Horseracing Board v. William Hill (2004)

Core Issue:

Whether investment in creating and maintaining a horse racing database gives automatic database rights.

Judgment (EU Court of Justice):

• “Database right” protects investment in obtaining and verifying data
• BUT NOT investment in creating the data itself

Principle Established:

There must be “substantial investment in collection,” not generation.

Relevance to Biodiversity Mapping:

• Field surveys of species = data creation (not protected under database right)
• Compiled biodiversity repositories (e.g., global species catalogs) MAY be protected if:
  • Significant effort spent collecting/cleaning data
• AI-generated predictions are not automatically protected as databases

3. Google LLC v. Oracle America Inc. (2021)

Core Issue:

Whether copying software code structure (Java APIs) constitutes fair use.

Judgment:

• Supreme Court ruled Google’s reuse was fair use
• Emphasized:
  • Functional nature of software interfaces
  • Transformative use in new platform (Android)

Principle Established:

• Software interfaces and functional elements get limited protection
• Interoperability and innovation may outweigh strict IP control

Relevance to Biodiversity Mapping:

• Ecological modeling APIs (e.g., species prediction libraries):
  • May not be strictly protected if reused for innovation
• Conservation platforms can legally reuse functional algorithmic structures
• Encourages open biodiversity AI ecosystems

4. Waymo LLC v. Uber Technologies Inc. (2017–2018)

Core Issue:

Theft of self-driving car trade secrets, including LiDAR-related algorithms.

Judgment:

• Case settled, but court found strong evidence of:
  • Misappropriation of confidential autonomous vehicle technology

Principle Established:

• Machine learning models, sensor fusion systems, and datasets are protectable as trade secrets
• Employee movement + data transfer is high-risk IP violation

Relevance to Biodiversity Mapping:

AI biodiversity platforms often include:

• Habitat prediction models
• Satellite image classifiers
• Species detection neural networks

These are strongly protectable as trade secrets if:

• Kept confidential
• Access-controlled
• Not publicly disclosed

5. SAS Institute Inc. v. World Programming Ltd. (2010–2012)

Core Issue:

Whether software functionality and programming language behavior can be copyrighted.

Judgment (UK + EU courts):

• Functionality and programming methods are NOT protected by copyright
• Only literal code expression is protected

Principle Established:

• “Idea vs expression” distinction is strict in software law

Relevance to Biodiversity Mapping:

• Ecological algorithms (e.g., species distribution logic) cannot be monopolized
• Competitors can replicate:
  • Model logic
  • Prediction methods
• BUT cannot copy:
  • Source code
  • UI design
  • Documentation verbatim

6. Feist Publications v. Rural Telephone Service (1991) (Extended Insight for AI Mapping)

(Reinforced due to importance in data-heavy systems)

• AI biodiversity systems often assume “big dataset = ownership”
• Feist rejects this:
  • Ownership requires originality, not accumulation
• This directly limits attempts to monopolize ecological data used in conservation AI

3. Synthesis: What This Means for Biodiversity Mapping IP

A. What CAN be protected

• AI model code for habitat prediction (copyright)
• Training pipelines (trade secrets)
• Curated biodiversity datasets with investment (database rights in EU-like regimes)
• Visual biodiversity maps (creative expression)
• Proprietary feature engineering methods (trade secrets)

B. What CANNOT be protected

• Raw biodiversity data (species counts, coordinates)
• Ecological facts and patterns in nature
• General ML ideas (e.g., “use CNN for satellite classification”)
• Functional algorithm logic without expression

4. Practical Legal Challenge in Conservation Tech

Algorithmic biodiversity mapping creates tension between:

• Open science goals (global ecological protection)
  vs
• Private IP incentives (funding AI conservation tools)

Courts consistently lean toward:

• Protecting expression + investment
• NOT protecting nature-derived facts or abstract ecological knowledge