Recognition systems evolve to reduce the risk and costs of making recognition errors. Two sources of recognition error include perceptual error (error arising from inability to discriminate between objects) and template error (error arising from using the wrong recognition template). We focus on how template error shapes host defense against avian brood parasites. Prior experiments in American coots (Fulica americana), a conspecific brood parasite, demonstrated how hosts learn to recognize brood parasitic chicks by using predictable patterns of hatching order of host and parasite eggs. Here, we use these results to quantify the benefit of chick rejection as well as the cost of template error, and we then use mathematical models to explore fitness payoffs of chick recognition from different template acquisition mechanisms. We find that fitness differences between mechanisms do not fully explain aspects of the learning mechanism, such as why coots reacquire their recognition template each year. Other constraints arising from mating systems and genetic mechanisms likely influence which learning mechanisms for parasitic chick recognition is optimal. Our approach could be expanded to explore how mechanisms of template acquisition influence other recognition systems, including parasitic chick recognition in other brood parasite hosts.

Field data collection

We studied the dynamics of brood parasitism in American coots in wetlands near Williams Lake, British Columbia in 1987-1990 (417 nests) and again from 2005-2008 (258 nests). We monitored nests every 1-4 days during egg laying, and new eggs were marked individually with indelible pen on each nest check. We detected brood parasitism when we found more than one egg laid in a nest per day and identified parasitic eggs using egg features like shape and color. The accuracy of these methods has been previously validated using genetic techniques (Lyon et al. 2002). We monitored nests daily during the 3-9 day hatching period. For analyses of hatching patterns, we used 63 nests for which both laying sequence and hatching sequence of parasites relative to hosts were known. Calculations of relative survival of hosts and parasites at control broods (i.e. naturally parasitized broods that were unmanipulated except for chick tagging) were based on 35 nests for which detailed censuses were conducted until the end of the parental care period.

For both control and experimental nests, we hatched chicks in captivity to assure complete accuracy in matching each chick to the egg it hatched from. We took eggs from nests at first sign of pipping, typically one or two days before the chicks hatched. We then hatched each egg inside an individual mesh pouch in an incubator (Hovabator 1602N, GQF Manufacturing, Savannah, GA). We returned the chicks to nests within 24 hours of hatching, after attaching color-coded nape tags that were individually unique at each brood (Arnold et al. 2011). Because of a high degree of hatching asynchrony, nests were never left with less than two eggs or chicks, and parents did not abandon the nest during this period.

We conducted censuses periodically for at least 20 days, and up to 35 days, after the last chick was returned to the nest. Brood censuses and behavioral observations were conducted at close range (10-40 m) from floating blinds equipped with camouflage coverings, where the individually distinct chick tags could be observed easily with binoculars. We determined survival by counting chicks that were seen in one of the last two censuses.

Cross-fostering experiment design

We conducted cross-fostering experiments to investigate the learning mechanism used in chick recognition (Shizuka & Lyon 2010). In the “Host First” experiment, the hosts were provided with their own offspring during the learning period, i.e., first day of hatching. Conversely, in the ‘Foreign First’ experiment, we provided the experimental hosts with foreign chicks (i.e., experimental parasitic chicks) on the first hatching day. In both treatments, on all days after the first hatching day we matched each host chick that hatched on a given day with a foreign chick of the same age. All foreign chicks used in a given experimental nest came from the same donor clutch so that all nests had chicks from only two sets of parents. Subsequent survival rates of chicks in experimental broods were assessed using the same protocol as control broods.

Survival Rate_Natural.xls
This is a worksheet that shows how we calculated the values presented in Table 1, which shows the expected and observed survival rates of hosts in natural (control) broods with parasitic chicks. The data are broken down into the two sampling periods: 1987-90 (collected by BEL) and 2005-08 (collected by DS and BEL). The formulas used to calculate summary data are preserved in the cells of the spreadsheet. This summary data is presented in lieu of the raw data since the data format for the two sampling periods were saved in different formats.

For each hatch order category, the yellow highlighted cells show the number of nests used, observed number of host chicks that survived vs. died, and number of all chicks (host + parasite) survived vs. died. The green highlighted cells show the observed proportion of chicks survived for each hatch order category. These values are used in Table 1 to calculate the benefit of rejection in natural broods. 

#####

HF-FF_Survival.xlsx
This is the raw data that can be used to calculate survival rates used in Table 2. The raw data is provided from the single year (2008) when all "Host First" and "Foreign First" experiments were conducted.