Estimate NestWatch Summary Dates • nestwatchR

The NestWatch Open Dataset holds a wealth of nesting data acquired from different sources. Due to participant omissions, differing protocols over time, and assimilation of other datasets, missing data fields are to be expected. For instance, a participant may forget to input the clutch size, but reliable clutch size information may exist in the nest visits section if the participant recorded numbers of eggs and chicks over multiple nest checks. The nw.est... family of functions help users estimate Clutch.Size, First.Lay.Date, Hatch.Date, and Fledge.Date.

Each of these four estimation functions are coded to generate estimates for missing values, when possible, based on the best data available for each nesting attempt. Though these functions aim to use the best available data, the precision of the resulting estimates cannot be guaranteed. As the analyst, you will need to make judgement calls in each step below based on your research goals and the biology of your focal species.

Estimate Clutch Size

The maximum clutch size is often missing in NestWatch nest summary data. nw.estclutchsize() takes a merged NestWatch dataframe as data =, and has an optional output = argument to rename the resulting dataframe.

This function looks for data contained in several columns to estimate the maximum clutch size for the nesting attempt. Because NestWatch participants do not usually visit each nest daily or record all data fields on every visit, the data needed to estimate clutch size may be spread over many columns in the nest visits table. The function looks, by Attempt.ID, in the nest visits and summary data and finds (1) the maximum egg count, (2) the unhatched eggs count, (3) the maximum young count (dead + alive), and (4) the number of fledglings recorded. From these summaries, the function calculates the maximum number (during each nest check) of observed (1) host eggs, (2) young (alive/dead) + unhatched eggs, and (3) fledged + unhatched eggs, to generate the maximum contents of the nest at each nest stage (lay/incubation, nestling, and fledging).

Depending on the data available, the certainty of the clutch size estimate may vary, or in some cases, may not be estimable. For instance, a successful nest with 4 eggs recorded on several checks and 4 fledged young observed would have high certainty that the clutch size was 4. However, if a nest was observed with 2 eggs on one check and a week later was found depredated, then the clutch size is not certain (more eggs could have been laid, but were not detected). In this case, the clutch size would be estimated as 2, but may have been higher depending on the exact (unobserved) failure date.

# Load wrens example data
wrens <- nestwatchR::wren_quickstart
length(unique(wrens$Attempt.ID))
# > [1] 7626    # total attempts in wrens

# How many attempts have missing clutch sizes?
empty_clutchsize <- wrens %>% filter(is.na(Clutch.Size))
length(unique(empty_clutchsize$Attempt.ID))
# > [1] 2228    # number of attempts with 'NA' for clutch.size

# Run estclutchsize
nw.estclutchsize(data = wrens, output = "estimated.clutch")
still_empty <- estimated.clutch %>% filter(is.na(Clutch.Size))
length(unique(still_empty$Attempt.ID))
# > [1] 376     # number of attempts with inestimable clutch size

Estimate First Lay Date

The date when the first egg was laid (first lay date) can be reliably estimated if the nest was visited during the egg-laying phase. Precision of estimates will decrease for nests found later in the nesting cycle. The function nw.estfirstlay() can help you identify which attempts have enough data to estimate lay dates. This function takes a merged NestWatch dataframe as data =, a user-provided average nest phenology dataframe (phenology =, see below for details), and has an optional output = argument to rename the resulting dataframe.

This function uses data spread across several data fields within the NestWatch dataset to estimate the lay date of the first egg of the attempt’s clutch. In order, the function determines if (1) the nest was visited during the egg-laying phase, (2) the attempt has a hatch date, and (3) if the nest has a fledge date. By checking in this order, the function estimates the lay date from the most proximal event date. CAUTION: If the nest was visited during laying, the estimated lay date will likely be more precise than if estimated from the fledge date because there is variation in incubation and nestling timeframes. By counting backwards in time from the most proximal event through the appropriate nest phases, we can estimate first lay dates. The timeframe between these events (laying, incubation, nestling, fledging), varies by species and may vary due to other variables like latitude. To account for this, this function takes the required phenology = argument, which is a user-provided dataframe with expected, average species-level nest phenology data (you could use mean, median, or mode number of days, whichever is most relevant to your study). In this example, both species typically lay one egg per day, but this could be adjusted to accommodate species with longer egg-laying intervals such as raptors, large waterfowl, etc., where Eggs.per.day can = 0.5 for an egg-laying interval of one egg every other day:

phenology <- data.frame(Species = c("bewwre", "carwre"),
                        Clutch.Size  = c(5, 4), # mean clutch size
                        Eggs.per.Day = c(1, 1), # typical laying frequency
                        Incubation   = c(16, 16), # mean incubation period
                        Nestling     = c(16, 13)) # mean nestling period

DT::datatable(phenology, options = list(paging = F, scrollX = T, searching  = F))

As in this example, you can input phenological data for multiple species to estimate first lay dates for each.

# Reuse `estimated.clutch` data from above
length(unique(estimated.clutch$Attempt.ID))
# > [1] 7626    # total attempts in estimated.clutch

# How many attempts have missing first lay dates?
empty_firstlay <- estimated.clutch %>% filter(is.na(First.Lay.Date))
length(unique(empty_firstlay$Attempt.ID))
# > [1] 3008    # number of nest attempts with 'NA' for first lay date

# Run estfirstlay
nw.estfirstlay(data = estimated.clutch, phenology = phenology, output = "estimated.lay")
still_empty <- estimated.lay %>% filter(is.na(First.Lay.Date))
length(unique(still_empty$Attempt.ID))
# > [1] 2392     # number of first lay dates not estimated

Estimate Hatch Date

For nesting attempts where young hatched, hatch dates may be estimable if first lay date or fledge date is known. The function nw.esthatch() takes a merged NestWatch dataframe as data =, the same user-provided, average nest phenology dataframe as above (phenology =, see above for details), and has an optional output = argument to rename the resulting dataframe.

This function checks, in order, if (1) the Young.Total were recorded or if the nest fledged (positive indication of hatching), (2) then if a hatch date is available, and finally (3) if a known (non-estimated) fledge date is available. If these requirements are met, the function uses the user-input nest phenology time frames to estimate hatch date date. Estimated dates are denoted with a 1 in data$Hatch.Date.Estimated.

Note that NestWatch does not have a data field for specific nestling age, but rather has a categorical nestling age score (i.e.”naked young”, “partially feathered young”). Due to the uncertainty in day-age of this score between species and possibly between observers, this function does not utilize these data.

# Reuse `estimated.lay` data from above
length(unique(estimated.lay$Attempt.ID))
# > [1] 7626    # total attempts in estimated.lay

# How many attempts have missing hatch dates?
empty_hatch <- estimated.lay %>% filter(is.na(Hatch.Date))
length(unique(empty_hatch$Attempt.ID))
# > [1] 3680    # number of nest attempts with 'NA' hatch dates

# Run esthatch
nw.esthatch(data = estimated.lay, phenology = phenology, output = "estimated.hatch")
still_empty <- estimated.hatch %>% filter(is.na(Hatch.Date))
length(unique(still_empty$Attempt.ID))
# > [1] 3207     # number of hatch dates not estimated

Estimate Fledge Date

For nesting attempts where young were known to have fledged, fledge dates may be estimable if hatch date or first lay date is known. The function nw.estfledge(), like those estimation functions above, takes a merged NestWatch dataframe as data =, the user-provided, average nest phenology dataframe from above (phenology =, see above for details), and has an optional output = argument to rename the resulting dataframe.

This function checks, in order, (1) if the nest fledged successfully, and (2) if a Hatch.Date was recorded or then (3) if a First.Lay.Date was recorded. If these requirements are met, the function uses user-input nest phenology time frames to estimate fledge date. Estimated dates are denoted with a 1 in data$Fledge.Date.Estimated.

CAUTION: Nestling periods can be quite variable, depending upon parental behavior, environmental conditions, food availability, etc. Use estimated fledge dates with caution.

# Reuse `estimated.hatch` data from above
length(unique(estimated.hatch$Attempt.ID))
# > [1] 7626    # total attempts in estimated.hatch

# How many attempts have missing fledge dates?
empty_fledge <- estimated.hatch %>% filter(is.na(Fledge.Date))
length(unique(empty_fledge$Attempt.ID))
# > [1] 3779    # number of nest attempts with 'NA' fledge dates

# Run estfledge
nw.estfledge(data = estimated.hatch, phenology = phenology, output = "estimated.fledge")
still_empty <- estimated.fledge %>% filter(is.na(Fledge.Date))
length(unique(still_empty$Attempt.ID))
# > [1] 3483     # number of fledge dates not estimated