(The author has carried out and published most of the connectivity modeling described here, with some of the earliest (e.g., for CA) and largest (e.g., for China) connectivity models. Over the same period, he has also collected and published wildlife connectivity data/evidence, including using camera traps, wildlife-vehicle collision, opportunistic and transect observation, animal sign, GPS-collars, and genetic approaches. The opinions here are based on this experience.)

Over the last 40 years an idea has percolated to the top of conservation planning suggesting that protecting biodiversity in a contested and changing landscape, and now with climate change, is best accomplished by designating and purchasing a wildland network of “core” or “reserve” areas connected by “corridors” or “linkages”. This idea grew from two unrelated sources: 1) the pessimistic perspective that conservation planning should be primarily based on limited funding, not based on ecological need, and 2) the growing availability of GIS mapping and modeling tools. Although prevalent and persistent, this hypothetical approach of building networks of reserves and linkages has survived both a lack of a priori evidence for its value and lack of post hoc evidence for effectiveness in representing the needs of wildlife. Stated more simply, there is no published, scientific evidence that wildlife in general preferentially use these networks, or if the networks were exclusively protected, could survive in just these areas. At the same time, a fan club of sorts has grown up around the approach, fueling a mini-industry of people who continue to develop and publish un-tested connectivity models for regions (e.g., San Diego County, Western Colorado), US states (e.g., AZ, CA, NM, OR, WA), and countries (e.g., China).

The (Fatal) Problems:

1. Except for a few terrestrial-dwelling species (e.g., caribou, wildebeest), most wildlife species don’t restrict themselves to walking along predictable paths (e.g., “linkages”), instead seeming to wander opportunistically in response to resource availability, climate, massive disturbance (e.g., wildfire), dispersal imperatives, mating drives, or exploration of new territory. Indeed, there is good reason to think that some random and thus unpredictable movement is highly adaptive (similar to salmon returning to non-natal streams).
2. Except species that are highly intolerant of human disturbance, most species do not restrict themselves conveniently to mapped “core” or “reserve” areas (e.g., Iverson et al., 2023). Even those with well-documented aversion to humans will stray into backyards, across streets, and through agricultural areas.
3. Almost all connectivity models, and certainly those large-extent models carried out by many well-known consulting scientists, are not developed based on multi-species wildlife movement or occurrence and after development are not evaluated using wildlife occurrence and movement data. Basically they don’t follow straightforward, long-standing guidelines for ecological models to verify, calibrate and validate before use (Rykiel, 1996). Despite this, they are used as “data” to inform important land acquisition and wildlife crossing decisions.
4. None of the issues in (1) to (3) are new. “There is sometimes a tendency for scientists and conservationists to be captivated by the theoretically interesting issues such as metapopulations and corridors, at the expense of addressing the less interesting and perhaps more basic ones such as the degradation of habitat within fragments. It is important to stress once again that no evidence  supports the proposition that corridors can mitigate the overall loss of habitat  (Harrison 1994, Fahrig 1998, Rosenberg et al. 1997)” (Harrison and Bruna, 1999).  Nevertheless, a culture has developed around the ideas of wildlife core areas, corridors, and linkages despite the continued absence of evidence and even after evaluations showing that hypothetical “linkages” and “corridors” don’t explain wildlife occurrence or movement (e.g., LaPoint et al., 2013; Iverson et al., 2023).
5. Many conservation planning approaches are driven by organizations and processes that rely on land sparing (acquisition) for income and as an assumed only way to practice conservation. However, for large-extent connectivity planning, there is good evidence that this approach is unaffordable (Shilling and Girvetz, 2007). In addition, there is no reason to think land-sharing can’t complement land-sparing for biodiversity conservation, especially with limited funding and political will (Kremen, 2015).
6. If wildlife generally (not a select one or two) don’t preferentially use core areas and linkages, then prioritizing these hypothetically-valuable areas over other areas may harm conservation and biodiversity protection. Similarly, if wildlife don’t use hypothetical linkages, then using these linkage model outputs to plan costly wildlife crossings over highways may lead to low-value (for wildlife) crossing structures and less connectivity than could be achieved.

The Solutions:

1.  One of the most important and long-awaited steps that hypothetical connectivity modeling should go through is validation of ecological relevance. In the largest published example of this type of evaluation, Iverson et al. (2023) found that most common connectivity modeling approaches (least-cost path, circuit-theory, disturbance cost surface) failed to generally predict wildlife occurrence or movement. There are many ways to define and frame the validation process, but one of the most straightforward is from Rykiel (1996).

Steps:

A.      Verification: “a demonstration that the modeling formalism is correct.” (internal functioning is good)

B.      Validation: “a demonstration that a model within its domain of applicability possesses a satisfactory range of accuracy consistent with the intended application of the model,” (performance of model in field)

C.      Calibration: “the estimation and adjustment of model parameters and constants to improve the agreement between model output and a data set.”

2.  If models are to be used, they should be based upon evidence of wildlife occurrence and movement (e.g., LaPoint, 2013). This is likely to result in predictions that pass validation tests (1) and that are more likely to contribute to wildlife conservation.

3.  Alternatives to using connectivity modeling to make decisions include: a) for individual species planning, resource selection function models, which appear to predict wildlife occurrence and movement well; and b) where available, using evidence of wildlife occurrence and movement. RSF modeling is improving and usually includes good model validation. There are many ways that wildlife data are being collected, with the primary barrier to their use being the retention (non-sharing) of these data in feudal-system data-vaults.

4.  Practitioners should consider land-sharing for connectivity conservation. This is especially true as landscape change in response to climate change. Land acquisitions for hypothetical linkages may get left behind as vegetation communities and disturbances shift. This has been true in the San Diego Multi-Species Habitat Plan area, where most of the reserves and “corridors” purchased over the last 20 years have burned at least once, with no nearby refugia for wildlife to flee to (Shilling, unpublished analyses).

Citations

Harrison, S. and Bruna, E., 1999. Habitat fragmentation and large‐scale conservation: what do we know for sure?. Ecography, 22(3), pp.225-232.

Iverson, A.R., Waetjen, D. and Shilling, F., 2024. Functional landscape connectivity for a select few: Linkages do not consistently predict wildlife movement or occupancy. Landscape and Urban Planning, 243, p.104953.

Kremen, C., 2015. Reframing the land‐sparing/land‐sharing debate for biodiversity conservation. Annals of the New York Academy of Sciences, 1355(1), pp.52-76.

LaPoint, S., Gallery, P., Wikelski, M. and Kays, R., 2013. Animal behavior, cost-based corridor models, and real corridors. Landscape ecology, 28, pp.1615-1630.

Rykiel, E.J., Jr. 1996. Testing ecological models: the meaning of validation. Ecological Modeling, 90:229-244.

Shilling, F. and Girvetz, E., 2007. Physical and financial barriers to implementing a nature reserve network in the Sierra Nevada, California, USA. Landscape and Urban Planning, 80(1-2), pp.165-172.