Sign Surveys for Estimating Trend of a Remnant Brown Bear Ursus arctos Population in Northern Spain

1. What they did

Brown bears in Spain’s Cantabrian Mountains were still declining in the early 1990s despite legal protection. The regional recovery plans called for monitoring the population with non-intrusive methods. Counting females with cubs, used for grizzly bears in Yellowstone, does not work well here: Cantabrian bears are mostly nocturnal, secretive and live in forest, so they are hard to see, and radio-collaring them is difficult.

The authors ran a 3-year pilot study (1990 to 1992) to test whether counting bear tracks and scats along survey routes (sign surveys) could detect changes in bear numbers over time.

• Routes. A network of 67 to 72 routes on foot trails and unimproved dirt roads, spread across the whole bear range.
• Segments. Each route was divided into numbered 1.6 km segments, the basic sampling unit. From what they knew of bear movements, they considered this long enough for neighbouring segments to give reasonably independent records.
• Timing. Surveys ran twice a year, in spring (mid-April, before livestock went up into the mountains) and autumn (mid-November, after livestock came down), to reduce trampling of sign. Each round was done in as short a time as possible, so sign from the same bear was less likely to be counted on more than one segment.
• Recording. Observers counted the bear tracks and scats they could positively identify on each segment. Scats were collected for diet analysis and tracks were measured to help tell individuals apart.
• Analysis. They calculated how many segments would be needed to detect a change in sign occurrence between two samples.

2. What they found

Survey effort

• About 1,500 km were surveyed each year (696 to 819 km per session), on 928 to 957 segments. Each session took on average 26.6 days.

Bear sign

• Over the three years, observers recorded 69 tracks and 52 scats. The total number of signs per year fell from 47 (1990) and 53 (1991) to 21 (1992).
• Tracks outnumbered scats in every session except one (autumn 1990, with 7 tracks and 20 scats).
• Sign was very scarce: an average of 0.06 sign per segment, and only 4% of segments had any sign (6 to 24 segments per survey, 16 to 37 per year).

Statistical power

• To detect a 20% decrease in sign occurrence (significance level 0.10, power 0.80), they would need 1,361 segments a year. They sampled well below that, so the survey design gave little power to detect real changes in bear numbers.
• Sign density was remarkably low, resembling bear track counts in the Montana study, even though the Cantabrian surveys ran twice a year and covered 10 times more segments a year.

3. What they concluded

• The pilot sampling scheme does not have enough statistical power to detect significant declines in this population. Monitoring trends in low-density carnivore populations with sign surveys is inherently difficult, because power drops as populations become smaller.
• Raising sign density estimates by adding replicate counts within each year looks like the best option. Pooling data over many years would take too long.
• Surveys are only practical about four months a year (April to May and October to November), because of livestock grazing for most of the year. A hypothetical scheme: 25 trained observers (for example district game wardens or volunteers) each walking about 32 km (20 segments) per session, about 2 days of fieldwork, covering 500 segments. Four monthly surveys a year would sample about 2,000 segments.
• For most of southern Europe’s remnant brown bear populations, agencies run sign surveys once or twice a year and record total tracks and scats per distance surveyed, without considering biases such as non-independent observations. The authors say this gives little useful information on population trends, and that questionnaires on bear occurrence give highly variable data, useful only on a crude spatial scale.
• They suggest a future strategy with two parts: (1) more replicate counts to lower variance and increase power, and (2) monitoring bear distribution and occurrence over time and space, to detect changes such as dispersal, recolonization, loss of peripheral range, or bears returning to good habitat.

4. How tracking was or was not used

The whole study was a tracks and signs survey. Observers walked the routes counting bear tracks and scats on each 1.6 km segment. Only signs that could be positively identified as bear were recorded. Presence on a segment meant at least one track or scat. Tracks were measured to help tell individuals apart, and scats were collected for diet analysis (neither result is reported in this paper). For the analysis, the sign on each segment was defined as the most common sign type found there. The authors note that sign surveys are affected by tracking conditions and by the animals’ seasonal behaviour, and they timed the surveys for good tracking conditions, when bears were active and before or after livestock trampled the trails. No trailing was reported.

5. Tracker or observer certification

Reported but not validated. Only “experienced field observers” took part in the surveys. Inexperienced observers could join after accompanying field personnel for two full survey sessions to gain experience in sign identification. No certification or test of their skill is described. Several people who helped run the surveys are thanked by name.