Resolving issues with plant and interaction terms
zedomel opened this issue · comments
Based on the dataset from Bergamo, P.J., Susin Streher, N., Traveset, A., Wolowski, M. and Sazima, M. (2020), Pollination outcomes reveal negative density-dependence coupled with interspecific facilitation among plants. Ecol Lett, 23: 129-139. https://doi.org/10.1111/ele.13415, I could identify some use cases that the actual version of the vocabulary cannot handle.
We don't have term for number of visits, and so, each visit have to be recorded separated (one visit - one interaction). To simplify it and allows any interaction to includes any number of visits I'm introducing a new term numberOfVisits (see #63).
It will simplifies data standardization process by avoid the duplication of records, but also (I'm not sure) we can use this term (numberOfVisits) as an indicator of single or multiple visits for others term in the vocabulary (like #23 and #36).
A simplified example of usage of the terms numberOfVisits, availableFlowerQuantity (#40), and dwc:organismQuantity:
Core: interactions.csv
Row Type: dwc:Event
| dwc:eventID | dwc:eventDate | dwc:month | dwc:samplingEffort |
|---|---|---|---|
| evt_1 | 2017-03-14 | march | 18 minutes |
| evt_2 | 2017-03-15 | march | 185 minutes |
| evt_3 | 2018-04-18 | april | 125 minutes |
Occurrences extension: occurrences.csv
Row Type: dwc:Occurrence
| dwc:eventID | dwc:occurrenceID | dwc:scientificName | dwc:organismQuantity | dwc:organismQuantityType |
|---|---|---|---|---|
| evt_1 | occ_1 | Pfaffia tuberosa | 1 | individuals |
| evt_1 | occ_2 | Ceratina asunciana | 1 | individuals |
| evt_2 | occ_3 | Lithrea brasiliensis | 1 | individuals |
| evt_2 | occ_4 | Apis mellifera | 2 | individuals |
| evt_3 | occ_5 | Lithrea brasiliensis | 3 | individuals |
| evt_3 | occ_5 | Trigona spinipes | 5 | individuals |
MeasurementOrFact extension: mof.csv
Row Type: dwc:MeasurementOrFact
| dwc:eventID | dwc:measurementID | dwc:measurementType | dwc:measurementValue |
|---|---|---|---|
| evt_1 | m_1 | numberOfVisits |
1 |
| evt_1 | m_2 | availableFlowerQuantity |
18 |
| evt_2 | m_3 | numberOfVisits |
5 |
| evt_2 | m_4 | availableFlowerQuantity |
22 |
| evt_3 | m_5 | numberOfVisits |
16 |
| evt_3 | m_6 | availableFlowerQuantity |
58 |
Thus, from the example above we know that the interaction evt_1 recorded in 2017-03-14 is an interaction between one individual of the species Pfaffia tuberosa and one individual of the species Ceratina asunciana. We also know that the C. asunciana individual visits one flower of the plant and that the plant had 18 flowers available to floral visistors.
Additionally, we know that the interaction evt_2 recorded in 2017-03-15 is an interaction between one individual of the species Lithrea brasiliensis and 2 individuals of the species Apis mellifera. We also know that these two individuals of A. mellifera visit 5 flowers of the plant and that the plant had 22 flowers available.
Similarly, we also know that the interaction evt_3 recorded in 2018-04-18 is an interaction between 3 individuals of the species Lithrea brasiliensis and 5 individuals from the species Trigona spinipes. We also know that the individuals of T. spinipes visit 16 flowers of the 3 plant individuals and the total number of flowers available considering all 3 plants were equals to 58.
With those terms we will be able to calculate things like visitation rate (numberOfVisits/availableFlowerQuantity/samplingEffort) but other metrics can also be derived.
Curious to know what others think about it.
It may also resolve some issues with terms like:
cospecificPollenGrainsQuantitySingleVisit(#2)cospecificPollenGrainsQuantity(#41)pollenTubesQuantitySingleVisit(#12)pollenTubesQuantity(#35)removedPollenGrainsQuantitySingleVisit(#23)removedPollenGrainsQuantity(#36)heterospecificPollenGrainsQuantity(#16)
Why the term fertilizedOvulesQuantity (#3) does not have a corresponding fertilizedOvulesQuantitySingleVisit?
I'm not sure if it make sense. Further discussions are needed to clarify it. But, the idea is to use numberOfVisits (#63) as an indicator of single (e.g. 1 or single) or multiple visits (>1 or multiple). If we can use numberOfVisits that way then we don't need two terms for each state (single and multiple).
We can not use both terms (single and multiple) together in the same interaction, since the we are considering one interaction = one flower visitation and then every interaction is always a single visit. On other hand, the same focal flower could still receive many other visitors (multiple visits), and then the these terms have to be used within plant occurrence instead of the interaction, or even a more complex model (aggregate multiple interactions by using dwc:parentEventID).
Similarly to the terms dwc:organismQuantity and dwc:individualCount in DwC the term numberOfVisits is not obligatory, and when not filled (empty) it should not be interpreted as 1 but as unknown number of visits.
I think that the addition of the term numberOfVisits and its usage together with the term dwc:organismQuantity and availableFlowersQuantity will bring more flexibility and simplicity to the adoption of the vocabulary.
Another question is if availableFlowerQuantity is sufficient, or do we need another term like flowerQuantity (number of observed flowers) since someone could not considering all available flowers to do focal observation of the interactions and recording measurements (e.g. conspecificPollenGrainsQuantity).
Does anyone have any comments ???
Another issue it may help to solve is related to fruitSet (#34), fruitMass (#10), seedSet(#37) and seedMass (#13)
The fruitSet is the most problematic term that we have in the entire vocabulary. First its definition is about a proportion. It is not a raw data, but something calculate between the total number of fruits / total number of flowers exposed to floral visitors:
Proportion of the flowers exposed to floral visitors that yielded fruits
Second, without the term numberOfVisits and considering each interaction as one visit, the fruitSet does not make sense, since any interaction is always about just one flower!!! The term numberOfVisits should provide the flexibility needed to use fruitSet and the definition can be improved to provided a absolute/raw value instead of a proportion:
The total number of flowers that set mature fruit
Additionally, this reference should be useful to define what we mean by fruit set:
The term “fruit set” has been used in the literature to refer to both initial and final fruit set. Initial fruit set occurs shortly after anthesis and is associated with the beginning of swelling of the ovary. Final fruit set refers to the number of fruits on a tree when the fruits and seeds are mature (Sedgley and Griffin, 1989). When a flower has been successfully pollinated, growth of the ovary is stimulated and floral parts such as stamens and petals usually wilt and abscise. Such changes, which characterize transformation of a flower into a young fruit, comprise initial fruit set.
Source: Stephen G. Pallardy, CHAPTER 4 - Reproductive Growth, Editor(s): Stephen G. Pallardy, Physiology of Woody Plants (Third Edition), Academic Press, 2008, Pages 87-106, ISBN 9780120887651, https://doi.org/10.1016/B978-012088765-1.50005-1
If the terminology associated to the fruit set concept is really a proportion, then my suggestion is to algo change the name of the term to something like numberOfFruits. The term seedSet is not a proportion and may be it is only a matter of definition here.
The terms seedSet and seedMass should be changed to accommodate the addition of the term numberOfVisists by including that they are about the total number of seeds. Then, from the total mass of the seeds and total number of seeds it is possible to calculate (among other thins) the average mass of a seed (totalMass/totalOfSeeds). The same applies to fruits.
@pjbergamo thank you reply. Please, see my comments bellow:
2.1) Single visit terms
Single visit measurements could be assigned to the interaction between one
animal species and one plant species. Usually, such measurements are
conducted when one wants to estimate the "effectiveness" of an animal
species as a pollinator of a plant species. In other words, how many pollen
grains an animal species deposits in a single visit (and consequently, how
many pollen tubes, number of fruits, seeds and so on). Such measurements
fit well our database. The problem is that they are quite rare because it
is very time consuming to conduct effectiveness approaches. I would say we
can keep those terms as they are very adequate for our database and we can
create all corresponding "Single visit" terms.
If these kind of data are rare, I don't see much evidence of demand to standardize such data. I can still capture for the REBIPP database, but we don't need to mint terms for that. At the end, it will only make things more complicate, since people will have to know that can't use single terms and multiple terms at same time. Otherwise, because it is related to just one interaction it is simpler than multiple terms to be used, but again it they are quite rare we can simplify the vocabulary and keep only multiple visits terms. Those terms (single visit) can be added later if we see sufficient evidence for create such terms.
2.2) Multiple visit terms
Multiple visit terms would make more sense if we treat them as plant traits
(as we are treating flower color, nectar, etc). This because they cannot be
assigned to the interaction between one animal species and one plant
species. They are likely the product of multiple interactions (multiple
animals and one plant species). I am comparing with the other plant traits
because we also do not measure all specific flowers in a plant that
interacted with the animal. We measure some flowers in that population to
gain insight about average (and variability) of plant traits that we
suspect are important in defining plant-pollinator interactions. Similarly,
we measure reproductive success in some flowers of that population to gain
insight about average (and variability) of the consequences of
plant-pollinator interactions. Depending on the study, interactions, traits
and reproductive success may be measured in the same plant individuals (but
never in the same flowers) - however, it is not always the case.
If we treat such terms as plant traits then we will lost any links with the interaction. We still have, for example, the number of conspecific pollen grains deposited by multiple visits*, but we will no be able to recover which were the interacting animal(s) that transported the pollen grains. Any plant trait is totally separated from the interaction, since they can exist without any interaction being recorded (e.g flowerColor, sexualSystem), in other words, they are independent of the interaction. On other hand, if the interaction information is required to the interpretation of such data then these terms have to be linked to the interactions some how. If it is fine to "lost" this information about the interacting animal, then it is fine to treat it as plant trait, otherwise, we need to elaborate a better solution/definition for those terms.
The way it is now, a dataset like this one:
| plant | animal | availableFlowersQuantity | numberOfVisitedFlowers | conspecificPollenGrainsQuantity | interactionType |
|---|---|---|---|---|---|
| Bertholletia excelsa | Bombus transversalis | 18 | 10 | 65 | visitsFlowersOf |
Should be interpreted as
an unknown number of individuals of the species Bombus transversalis visited 10 flowers of an unknown number of individuals of the species Bertholletia excelsa*, which has a total of 18 flowers available (fully open), and the 10 visits are responsible for the deposition of 65 conspecific pollen grains.
On other hand if we treat conspecificPollenGrainsQuantity as a plant trait then the interpretation should be slightly different:
an unknown number of individuals of the species Bombus transversalis visited 10 flowers of an unknown number of individuals of the species Bertholletia excelsa*, which has a total of 18 flowers available (fully open), and it was found a total 65 conspecific pollen grains deposited in the flowers' stigmas of the plant individuals by multiple visitors (not only this 10 visits)
Is it right? Again, if it is not the right interpretation for the term conspecificPollenGrainsQuantitythen we need to revise the term, and think in a solution that fits it usage and can be used with DwC.
2.3) Fruit set (multiple visits)
Fruit set is not related to the number of flowers exposed to animal
visitors. Usually, fruit set is measured by marking a subset of the flowers
available (it is very hard to mark all available flowers, especially in
trees or shrubs with multiple branches) and, from these subset of flowers
one will then count how many set fruit
If fruitSet is not related to the availableFlowerQuantity, perhaps we need to find a way to capture the number of marked flowers or something similar. But the same problem of conspecificPollenGrainsQuantity applies here, since fruitSet is related to multiple interactions. Then, if fruitSet is defined as the total number of flowers that set mature fruits then we need to know the total number of flowers (marked flowers).
2.4) fruitMass, seedSet, seedMass
These measurements are not proportions and should be treated as
cospecificPollenGrains (and the other pollen-related reproductive success
variables: pollen tubes, fertilized ovules and so on).
This are less concern terms. Since they are defined to be single visit they can be document together with the interaction.